{"id":72684,"date":"2026-02-06T11:50:01","date_gmt":"2026-02-06T03:50:01","guid":{"rendered":"https:\/\/www.wsisp.com\/helps\/72684.html"},"modified":"2026-02-06T11:50:01","modified_gmt":"2026-02-06T03:50:01","slug":"%e3%80%90%e5%ae%8c%e6%95%b4%e6%ba%90%e7%a0%81%e6%95%b0%e6%8d%ae%e9%9b%86%e9%83%a8%e7%bd%b2%e6%95%99%e7%a8%8b%e3%80%91%e9%be%8b%e9%bd%bf%e7%89%99%e9%bd%bf%e7%97%85%e5%8f%98%e5%9b%be%e5%83%8f%e5%88%86","status":"publish","type":"post","link":"https:\/\/www.wsisp.com\/helps\/72684.html","title":{"rendered":"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b"},"content":{"rendered":"

\u80cc\u666f\u610f\u4e49<\/h2>\n

\u9f8b\u9f7f\u4f5c\u4e3a\u4e00\u79cd\u5e38\u89c1\u7684\u53e3\u8154\u75be\u75c5，\u5f71\u54cd\u7740\u5168\u7403\u6570\u4ebf\u4eba\u7684\u53e3\u8154\u5065\u5eb7\u3002\u6839\u636e\u4e16\u754c\u536b\u751f\u7ec4\u7ec7\u7684\u7edf\u8ba1，\u9f8b\u9f7f\u7684\u53d1\u75c5\u7387\u5728\u513f\u7ae5\u548c\u9752\u5c11\u5e74\u4e2d\u5c24\u4e3a\u663e\u8457，\u4e14\u968f\u7740\u5e74\u9f84\u7684\u589e\u957f，\u6210\u4eba\u548c\u8001\u5e74\u4eba\u4e5f\u9762\u4e34\u7740\u9f8b\u9f7f\u7684\u5a01\u80c1\u3002\u9f8b\u9f7f\u4e0d\u4ec5\u4f1a\u5bfc\u81f4\u75bc\u75db\u548c\u4e0d\u9002，\u8fd8\u53ef\u80fd\u5f15\u53d1\u66f4\u4e3a\u4e25\u91cd\u7684\u53e3\u8154\u5e76\u53d1\u75c7，\u5982\u7259\u9ad3\u708e\u3001\u6839\u5c16\u5468\u56f4\u708e\u7b49\u3002\u56e0\u6b64，\u65e9\u671f\u7684\u9f8b\u9f7f\u68c0\u6d4b\u4e0e\u8bca\u65ad\u663e\u5f97\u5c24\u4e3a\u91cd\u8981\u3002\u4f20\u7edf\u7684\u9f8b\u9f7f\u8bca\u65ad\u65b9\u6cd5\u4e3b\u8981\u4f9d\u8d56\u4e8e\u533b\u751f\u7684\u89c6\u89c9\u68c0\u67e5\u548cX\u5149\u5f71\u50cf，\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u4ec5\u8017\u65f6，\u800c\u4e14\u5bb9\u6613\u53d7\u5230\u4e3b\u89c2\u56e0\u7d20\u7684\u5f71\u54cd，\u5bfc\u81f4\u6f0f\u8bca\u6216\u8bef\u8bca\u7684\u60c5\u51b5\u3002<\/p>\n

\u968f\u7740\u8ba1\u7b97\u673a\u89c6\u89c9\u548c\u6df1\u5ea6\u5b66\u4e60\u6280\u672f\u7684\u8fc5\u731b\u53d1\u5c55，\u57fa\u4e8e\u56fe\u50cf\u5904\u7406\u7684\u81ea\u52a8\u5316\u8bca\u65ad\u7cfb\u7edf\u9010\u6e10\u6210\u4e3a\u7814\u7a76\u7684\u70ed\u70b9\u3002YOLO（You Only Look Once）\u7cfb\u5217\u6a21\u578b\u56e0\u5176\u9ad8\u6548\u7684\u5b9e\u65f6\u76ee\u6807\u68c0\u6d4b\u80fd\u529b\u800c\u53d7\u5230\u5e7f\u6cdb\u5173\u6ce8\u3002YOLOv8\u4f5c\u4e3a\u8be5\u7cfb\u5217\u7684\u6700\u65b0\u7248\u672c，\u5177\u5907\u66f4\u5f3a\u7684\u7279\u5f81\u63d0\u53d6\u80fd\u529b\u548c\u66f4\u9ad8\u7684\u68c0\u6d4b\u7cbe\u5ea6，\u9002\u5408\u7528\u4e8e\u590d\u6742\u7684\u533b\u5b66\u56fe\u50cf\u5206\u6790\u4efb\u52a1\u3002\u901a\u8fc7\u5bf9YOLOv8\u8fdb\u884c\u6539\u8fdb，\u53ef\u4ee5\u8fdb\u4e00\u6b65\u63d0\u5347\u5176\u5728\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u4e2d\u7684\u5e94\u7528\u6548\u679c，\u4e3a\u4e34\u5e8a\u63d0\u4f9b\u66f4\u4e3a\u51c6\u786e\u7684\u8f85\u52a9\u8bca\u65ad\u5de5\u5177\u3002<\/p>\n

\u672c\u7814\u7a76\u65e8\u5728\u57fa\u4e8e\u6539\u8fdb\u7684YOLOv8\u6a21\u578b，\u6784\u5efa\u4e00\u4e2a\u9ad8\u6548\u7684\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u3002\u6211\u4eec\u5c06\u5229\u7528\u4e00\u4e2a\u5305\u542b5100\u5f20\u56fe\u50cf\u7684\u9f8b\u9f7f\u5206\u5272\u6570\u636e\u96c6，\u8be5\u6570\u636e\u96c6\u6db5\u76d6\u4e8623\u4e2a\u7c7b\u522b，\u5305\u62ec\u5404\u79cd\u7c7b\u578b\u7684\u7259\u9f7f\u75c5\u53d8，\u5982\u9f8b\u9f7f\u3001\u6839\u7ba1\u6cbb\u7597\u3001\u7259\u51a0\u3001\u7259\u6865\u7b49\u3002\u8fd9\u4e9b\u7c7b\u522b\u7684\u591a\u6837\u6027\u4e3a\u6a21\u578b\u7684\u8bad\u7ec3\u63d0\u4f9b\u4e86\u4e30\u5bcc\u7684\u6837\u672c，\u6709\u52a9\u4e8e\u63d0\u9ad8\u6a21\u578b\u7684\u6cdb\u5316\u80fd\u529b\u548c\u51c6\u786e\u6027\u3002\u6b64\u5916，\u6570\u636e\u96c6\u4e2d\u4e0d\u540c\u75c5\u53d8\u7684\u6807\u6ce8\u4fe1\u606f\u5c06\u4e3a\u6a21\u578b\u7684\u5b66\u4e60\u63d0\u4f9b\u6e05\u6670\u7684\u76ee\u6807，\u4fc3\u8fdb\u5176\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u7684\u8868\u73b0\u3002<\/p>\n

\u672c\u7814\u7a76\u7684\u610f\u4e49\u4e0d\u4ec5\u5728\u4e8e\u63d0\u5347\u9f8b\u9f7f\u7684\u81ea\u52a8\u5316\u68c0\u6d4b\u80fd\u529b，\u8fd8\u5728\u4e8e\u63a8\u52a8\u53e3\u8154\u533b\u5b66\u4e0e\u4eba\u5de5\u667a\u80fd\u6280\u672f\u7684\u6df1\u5ea6\u878d\u5408\u3002\u901a\u8fc7\u6784\u5efa\u4e00\u4e2a\u9ad8\u6548\u7684\u56fe\u50cf\u5206\u5272\u7cfb\u7edf，\u80fd\u591f\u4e3a\u7259\u79d1\u533b\u751f\u63d0\u4f9b\u66f4\u4e3a\u7cbe\u51c6\u7684\u75c5\u53d8\u8bc6\u522b\u5de5\u5177，\u5e2e\u52a9\u5176\u5728\u4e34\u5e8a\u5b9e\u8df5\u4e2d\u505a\u51fa\u66f4\u4e3a\u79d1\u5b66\u7684\u51b3\u7b56\u3002\u6b64\u5916，\u8be5\u7cfb\u7edf\u7684\u6210\u529f\u5e94\u7528\u4e5f\u5c06\u4e3a\u5176\u4ed6\u53e3\u8154\u75be\u75c5\u7684\u81ea\u52a8\u5316\u68c0\u6d4b\u63d0\u4f9b\u501f\u9274，\u63a8\u52a8\u6574\u4e2a\u53e3\u8154\u533b\u5b66\u9886\u57df\u7684\u6280\u672f\u8fdb\u6b65\u3002<\/p>\n

\u7efc\u4e0a\u6240\u8ff0，\u57fa\u4e8e\u6539\u8fdbYOLOv8\u7684\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u7684\u7814\u7a76，\u4e0d\u4ec5\u5177\u6709\u91cd\u8981\u7684\u7406\u8bba\u4ef7\u503c，\u8fd8\u6709\u52a9\u4e8e\u89e3\u51b3\u5b9e\u9645\u4e34\u5e8a\u4e2d\u7684\u95ee\u9898，\u4e3a\u63d0\u9ad8\u53e3\u8154\u5065\u5eb7\u6c34\u5e73\u505a\u51fa\u8d21\u732e\u3002\u968f\u7740\u7814\u7a76\u7684\u6df1\u5165，\u6211\u4eec\u671f\u5f85\u8be5\u7cfb\u7edf\u80fd\u591f\u5728\u672a\u6765\u7684\u53e3\u8154\u533b\u7597\u4e2d\u53d1\u6325\u91cd\u8981\u4f5c\u7528，\u52a9\u529b\u5b9e\u73b0\u66f4\u4e3a\u7cbe\u51c6\u548c\u9ad8\u6548\u7684\u53e3\u8154\u75be\u75c5\u7ba1\u7406\u3002<\/p>\n

\u56fe\u7247\u6548\u679c<\/h2>\n

\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

\u6570\u636e\u96c6\u4fe1\u606f<\/h2>\n

\u5728\u73b0\u4ee3\u7259\u79d1\u533b\u5b66\u4e2d，\u9f8b\u9f7f\u7684\u65e9\u671f\u8bca\u65ad\u4e0e\u6cbb\u7597\u81f3\u5173\u91cd\u8981，\u800c\u56fe\u50cf\u5206\u5272\u6280\u672f\u5728\u8fd9\u4e00\u9886\u57df\u7684\u5e94\u7528\u4e3a\u7259\u79d1\u533b\u751f\u63d0\u4f9b\u4e86\u66f4\u4e3a\u7cbe\u786e\u7684\u5de5\u5177\u3002\u4e3a\u6b64，\u6211\u4eec\u6784\u5efa\u4e86\u4e00\u4e2a\u540d\u4e3a\u201ctooth segmentation\u201d\u7684\u6570\u636e\u96c6，\u65e8\u5728\u8bad\u7ec3\u6539\u8fdb\u7248\u7684YOLOv8-seg\u6a21\u578b，\u4ee5\u5b9e\u73b0\u5bf9\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u7684\u9ad8\u6548\u5206\u5272\u3002\u8be5\u6570\u636e\u96c6\u4e13\u6ce8\u4e8e\u516b\u79cd\u4e0d\u540c\u7684\u7259\u9f7f\u75c5\u53d8\u7c7b\u578b，\u6db5\u76d6\u4e86\u4ece\u9f8b\u9f7f\u5230\u5404\u79cd\u4fee\u590d\u4f53\u7684\u591a\u6837\u6027，\u786e\u4fdd\u6a21\u578b\u80fd\u591f\u8bc6\u522b\u548c\u5206\u5272\u8fd9\u4e9b\u5173\u952e\u7279\u5f81\u3002<\/p>\n

\u201ctooth segmentation\u201d\u6570\u636e\u96c6\u5305\u542b\u516b\u4e2a\u7c7b\u522b，\u5206\u522b\u4e3a：\u9f8b\u9f7f（Caries）\u3001\u7259\u51a0\u4e0e\u6865（Crown-Bridge）\u3001\u586b\u5145\u7269（Filling）\u3001\u690d\u7259（Implant）\u3001\u540e\u87ba\u9489（Post-screw）\u3001\u6839\u7ba1\u586b\u5145（Root canal Obturation）\u3001\u7259\u51a0（crown）\u4ee5\u53ca\u518d\u690d\u4f53（implant）\u3002\u8fd9\u4e9b\u7c7b\u522b\u4e0d\u4ec5\u4ee3\u8868\u4e86\u7259\u9f7f\u7684\u4e0d\u540c\u75c5\u53d8\u72b6\u6001，\u8fd8\u6db5\u76d6\u4e86\u7259\u79d1\u6cbb\u7597\u4e2d\u5e38\u89c1\u7684\u4fee\u590d\u548c\u91cd\u5efa\u8fc7\u7a0b\u3002\u8fd9\u79cd\u591a\u6837\u6027\u4f7f\u5f97\u6570\u636e\u96c6\u5728\u8bad\u7ec3\u8fc7\u7a0b\u4e2d\u80fd\u591f\u6709\u6548\u5730\u63d0\u9ad8\u6a21\u578b\u7684\u6cdb\u5316\u80fd\u529b，\u5e2e\u52a9\u5176\u5728\u5b9e\u9645\u5e94\u7528\u4e2d\u66f4\u597d\u5730\u9002\u5e94\u4e0d\u540c\u7684\u75c5\u4f8b\u3002<\/p>\n

\u5728\u6570\u636e\u96c6\u7684\u6784\u5efa\u8fc7\u7a0b\u4e2d，\u6211\u4eec\u6ce8\u91cd\u56fe\u50cf\u7684\u8d28\u91cf\u548c\u6807\u6ce8\u7684\u51c6\u786e\u6027\u3002\u6bcf\u4e00\u5f20\u56fe\u50cf\u90fd\u7ecf\u8fc7\u4e13\u4e1a\u7259\u79d1\u533b\u751f\u7684\u5ba1\u6838\u4e0e\u6807\u6ce8，\u786e\u4fdd\u6bcf\u4e2a\u7c7b\u522b\u7684\u7279\u5f81\u90fd\u88ab\u51c6\u786e\u5730\u8bc6\u522b\u548c\u6807\u8bb0\u3002\u8fd9\u4e00\u8fc7\u7a0b\u4e0d\u4ec5\u63d0\u9ad8\u4e86\u6570\u636e\u96c6\u7684\u53ef\u9760\u6027，\u4e5f\u4e3a\u540e\u7eed\u7684\u6a21\u578b\u8bad\u7ec3\u63d0\u4f9b\u4e86\u575a\u5b9e\u7684\u57fa\u7840\u3002\u6570\u636e\u96c6\u4e2d\u5305\u542b\u7684\u56fe\u50cf\u6765\u6e90\u4e8e\u771f\u5b9e\u7684\u4e34\u5e8a\u75c5\u4f8b，\u6db5\u76d6\u4e86\u4e0d\u540c\u5e74\u9f84\u6bb5\u548c\u6027\u522b\u7684\u60a3\u8005，\u786e\u4fdd\u4e86\u6837\u672c\u7684\u591a\u6837\u6027\u548c\u4ee3\u8868\u6027\u3002<\/p>\n

\u4e3a\u4e86\u8fdb\u4e00\u6b65\u63d0\u5347\u6a21\u578b\u7684\u6027\u80fd，\u6211\u4eec\u5728\u6570\u636e\u96c6\u4e2d\u5f15\u5165\u4e86\u6570\u636e\u589e\u5f3a\u6280\u672f\u3002\u8fd9\u4e9b\u6280\u672f\u5305\u62ec\u65cb\u8f6c\u3001\u7f29\u653e\u3001\u7ffb\u8f6c\u548c\u4eae\u5ea6\u8c03\u6574\u7b49，\u65e8\u5728\u589e\u52a0\u8bad\u7ec3\u6837\u672c\u7684\u591a\u6837\u6027，\u51cf\u5c11\u6a21\u578b\u5bf9\u7279\u5b9a\u56fe\u50cf\u7279\u5f81\u7684\u8fc7\u62df\u5408\u3002\u901a\u8fc7\u8fd9\u4e9b\u589e\u5f3a\u624b\u6bb5，\u6211\u4eec\u5e0c\u671b\u6a21\u578b\u80fd\u591f\u5728\u9762\u5bf9\u4e0d\u540c\u7684\u56fe\u50cf\u53d8\u6362\u65f6，\u4f9d\u7136\u4fdd\u6301\u8f83\u9ad8\u7684\u5206\u5272\u7cbe\u5ea6\u3002<\/p>\n

\u6b64\u5916，\u6570\u636e\u96c6\u7684\u6784\u5efa\u8fd8\u8003\u8651\u5230\u4e86\u6a21\u578b\u7684\u5b9e\u9645\u5e94\u7528\u573a\u666f\u3002\u968f\u7740\u7259\u79d1\u5f71\u50cf\u6280\u672f\u7684\u4e0d\u65ad\u8fdb\u6b65，\u7259\u79d1\u533b\u751f\u9762\u4e34\u7740\u8d8a\u6765\u8d8a\u591a\u7684\u5f71\u50cf\u6570\u636e，\u5982\u4f55\u5feb\u901f\u3001\u51c6\u786e\u5730\u8bc6\u522b\u548c\u5206\u5272\u51fa\u75c5\u53d8\u533a\u57df\u6210\u4e3a\u4e86\u4e00\u4e2a\u91cd\u8981\u7684\u6311\u6218\u3002\u56e0\u6b64，\u8bad\u7ec3\u51fa\u7684YOLOv8-seg\u6a21\u578b\u4e0d\u4ec5\u9700\u8981\u5177\u5907\u9ad8\u6548\u7684\u5904\u7406\u80fd\u529b，\u8fd8\u9700\u5728\u5b9e\u65f6\u5e94\u7528\u4e2d\u4fdd\u6301\u826f\u597d\u7684\u54cd\u5e94\u901f\u5ea6\u3002\u4e3a\u6b64，\u6211\u4eec\u5728\u6570\u636e\u96c6\u4e2d\u8bbe\u8ba1\u4e86\u591a\u79cd\u4e0d\u540c\u5149\u7167\u548c\u80cc\u666f\u6761\u4ef6\u4e0b\u7684\u56fe\u50cf，\u4ee5\u63d0\u9ad8\u6a21\u578b\u5728\u590d\u6742\u73af\u5883\u4e2d\u7684\u9002\u5e94\u6027\u3002<\/p>\n

\u603b\u4e4b，\u201ctooth segmentation\u201d\u6570\u636e\u96c6\u7684\u6784\u5efa\u4e3a\u6539\u8fdbYOLOv8-seg\u6a21\u578b\u7684\u8bad\u7ec3\u63d0\u4f9b\u4e86\u575a\u5b9e\u7684\u57fa\u7840，\u65e8\u5728\u63a8\u52a8\u7259\u79d1\u5f71\u50cf\u5206\u6790\u6280\u672f\u7684\u53d1\u5c55\u3002\u901a\u8fc7\u5bf9\u516b\u79cd\u7259\u9f7f\u75c5\u53d8\u7684\u7cbe\u786e\u5206\u5272，\u671f\u671b\u80fd\u591f\u4e3a\u7259\u79d1\u533b\u751f\u63d0\u4f9b\u66f4\u4e3a\u9ad8\u6548\u7684\u8f85\u52a9\u5de5\u5177，\u63d0\u5347\u9f8b\u9f7f\u53ca\u5176\u4ed6\u7259\u9f7f\u75c5\u53d8\u7684\u8bca\u65ad\u548c\u6cbb\u7597\u6548\u7387\u3002\u968f\u7740\u6570\u636e\u96c6\u7684\u4e0d\u65ad\u5b8c\u5584\u548c\u6a21\u578b\u7684\u6301\u7eed\u4f18\u5316，\u6211\u4eec\u76f8\u4fe1\u8fd9\u4e00\u7814\u7a76\u5c06\u4e3a\u7259\u79d1\u533b\u5b66\u7684\u672a\u6765\u53d1\u5c55\u5f00\u8f9f\u65b0\u7684\u53ef\u80fd\u6027\u3002
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"
\n\"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

\u6838\u5fc3\u4ee3\u7801<\/h2>\n

```\u4ee5\u4e0b\u662f\u5bf9\u7ed9\u5b9a\u4ee3\u7801\u7684\u6838\u5fc3\u90e8\u5206\u8fdb\u884c\u5206\u6790\u548c\u6ce8\u91ca\u7684\u7ed3\u679c\u3002\u4ee3\u7801\u7684\u4e3b\u8981\u529f\u80fd\u662f\u5b9e\u73b0YOLOv8\u6a21\u578b\u7684\u4e0d\u540c\u5934\u90e8\u6a21\u5757，\u5305\u62ec\u68c0\u6d4b\u3001\u5206\u5272\u3001\u59ff\u6001\u4f30\u8ba1\u548c\u5206\u7c7b\u7b49\u3002\u6211\u4eec\u5c06\u4fdd\u7559\u6838\u5fc3\u7684Detect\u7c7b\u53ca\u5176\u5b50\u7c7b，\u5e76\u5bf9\u5176\u8fdb\u884c\u8be6\u7ec6\u6ce8\u91ca\u3002<\/p>\n

import<\/span> torch
\nimport<\/span> torch.<\/span>nn as<\/span> nn
\nfrom<\/span> .<\/span>block import<\/span> DFL,<\/span> Proto
\nfrom<\/span> .<\/span>conv import<\/span> Conv
\nfrom<\/span> .<\/span>transformer import<\/span> MLP<\/p>\n

class<\/span> Detect<\/span>(<\/span>nn.<\/span>Module)<\/span>:<\/span>
\n """YOLOv8\u68c0\u6d4b\u5934，\u7528\u4e8e\u76ee\u6807\u68c0\u6d4b\u6a21\u578b\u3002"""<\/span><\/p>\n

def<\/span> __init__<\/span>(<\/span>self,<\/span> nc=<\/span>80<\/span>,<\/span> ch=<\/span>(<\/span>)<\/span>)<\/span>:<\/span>
\n """\u521d\u59cb\u5316YOLOv8\u68c0\u6d4b\u5c42，\u6307\u5b9a\u7c7b\u522b\u6570\u548c\u901a\u9053\u6570\u3002<\/p>\n

Args:
\n nc (int): \u7c7b\u522b\u6570，\u9ed8\u8ba4\u4e3a80\u3002
\n ch (tuple): \u8f93\u5165\u901a\u9053\u6570\u7684\u5143\u7ec4\u3002
\n """<\/span>
\n super<\/span>(<\/span>)<\/span>.<\/span>__init__(<\/span>)<\/span>
\n self.<\/span>nc =<\/span> nc # \u7c7b\u522b\u6570<\/span>
\n self.<\/span>nl =<\/span> len<\/span>(<\/span>ch)<\/span> # \u68c0\u6d4b\u5c42\u7684\u6570\u91cf<\/span>
\n self.<\/span>reg_max =<\/span> 16<\/span> # DFL\u901a\u9053\u6570<\/span>
\n self.<\/span>no =<\/span> nc +<\/span> self.<\/span>reg_max *<\/span> 4<\/span> # \u6bcf\u4e2a\u951a\u70b9\u7684\u8f93\u51fa\u6570\u91cf<\/span>
\n self.<\/span>stride =<\/span> torch.<\/span>zeros(<\/span>self.<\/span>nl)<\/span> # \u5728\u6784\u5efa\u8fc7\u7a0b\u4e2d\u8ba1\u7b97\u7684\u6b65\u5e45<\/span>
\n c2,<\/span> c3 =<\/span> max<\/span>(<\/span>(<\/span>16<\/span>,<\/span> ch[<\/span>0<\/span>]<\/span> \/\/<\/span> 4<\/span>,<\/span> self.<\/span>reg_max *<\/span> 4<\/span>)<\/span>)<\/span>,<\/span> max<\/span>(<\/span>ch[<\/span>0<\/span>]<\/span>,<\/span> min<\/span>(<\/span>self.<\/span>nc,<\/span> 100<\/span>)<\/span>)<\/span> # \u901a\u9053\u6570<\/span><\/p>\n

# \u5b9a\u4e49\u4e24\u4e2a\u5377\u79ef\u6a21\u5757 cv2 \u548c cv3<\/span>
\n self.<\/span>cv2 =<\/span> nn.<\/span>ModuleList(<\/span>
\n nn.<\/span>Sequential(<\/span>Conv(<\/span>x,<\/span> c2,<\/span> 3<\/span>)<\/span>,<\/span> Conv(<\/span>c2,<\/span> c2,<\/span> 3<\/span>)<\/span>,<\/span> nn.<\/span>Conv2d(<\/span>c2,<\/span> 4<\/span> *<\/span> self.<\/span>reg_max,<\/span> 1<\/span>)<\/span>)<\/span> for<\/span> x in<\/span> ch)<\/span>
\n self.<\/span>cv3 =<\/span> nn.<\/span>ModuleList(<\/span>nn.<\/span>Sequential(<\/span>Conv(<\/span>x,<\/span> c3,<\/span> 3<\/span>)<\/span>,<\/span> Conv(<\/span>c3,<\/span> c3,<\/span> 3<\/span>)<\/span>,<\/span> nn.<\/span>Conv2d(<\/span>c3,<\/span> self.<\/span>nc,<\/span> 1<\/span>)<\/span>)<\/span> for<\/span> x in<\/span> ch)<\/span><\/p>\n

# DFL\u6a21\u5757，\u5982\u679creg_max\u5927\u4e8e1\u5219\u4f7f\u7528DFL，\u5426\u5219\u4f7f\u7528\u8eab\u4efd\u6620\u5c04<\/span>
\n self.<\/span>dfl =<\/span> DFL(<\/span>self.<\/span>reg_max)<\/span> if<\/span> self.<\/span>reg_max ><\/span> 1<\/span> else<\/span> nn.<\/span>Identity(<\/span>)<\/span><\/p>\n

def<\/span> forward<\/span>(<\/span>self,<\/span> x)<\/span>:<\/span>
\n """\u8fde\u63a5\u5e76\u8fd4\u56de\u9884\u6d4b\u7684\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6982\u7387\u3002<\/p>\n

Args:
\n x (list): \u8f93\u5165\u7279\u5f81\u56fe\u7684\u5217\u8868\u3002<\/p>\n

Returns:
\n Tensor: \u9884\u6d4b\u7684\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6982\u7387\u3002
\n """<\/span>
\n shape =<\/span> x[<\/span>0<\/span>]<\/span>.<\/span>shape # BCHW<\/span>
\n for<\/span> i in<\/span> range<\/span>(<\/span>self.<\/span>nl)<\/span>:<\/span>
\n # \u5bf9\u6bcf\u4e2a\u68c0\u6d4b\u5c42\u7684\u8f93\u5165\u8fdb\u884c\u5377\u79ef\u64cd\u4f5c\u5e76\u8fde\u63a5<\/span>
\n x[<\/span>i]<\/span> =<\/span> torch.<\/span>cat(<\/span>(<\/span>self.<\/span>cv2[<\/span>i]<\/span>(<\/span>x[<\/span>i]<\/span>)<\/span>,<\/span> self.<\/span>cv3[<\/span>i]<\/span>(<\/span>x[<\/span>i]<\/span>)<\/span>)<\/span>,<\/span> 1<\/span>)<\/span><\/p>\n

# \u5982\u679c\u5728\u8bad\u7ec3\u6a21\u5f0f\u4e0b，\u76f4\u63a5\u8fd4\u56de\u5904\u7406\u540e\u7684\u7279\u5f81<\/span>
\n if<\/span> self.<\/span>training:<\/span>
\n return<\/span> x<\/p>\n

# \u52a8\u6001\u751f\u6210\u951a\u70b9\u548c\u6b65\u5e45<\/span>
\n self.<\/span>anchors,<\/span> self.<\/span>strides =<\/span> (<\/span>x.<\/span>transpose(<\/span>0<\/span>,<\/span> 1<\/span>)<\/span> for<\/span> x in<\/span> make_anchors(<\/span>x,<\/span> self.<\/span>stride,<\/span> 0.5<\/span>)<\/span>)<\/span>
\n self.<\/span>shape =<\/span> shape<\/p>\n

# \u5c06\u6240\u6709\u68c0\u6d4b\u5c42\u7684\u8f93\u51fa\u8fde\u63a5\u6210\u4e00\u4e2a\u5f20\u91cf<\/span>
\n x_cat =<\/span> torch.<\/span>cat(<\/span>[<\/span>xi.<\/span>view(<\/span>shape[<\/span>0<\/span>]<\/span>,<\/span> self.<\/span>no,<\/span> –<\/span>1<\/span>)<\/span> for<\/span> xi in<\/span> x]<\/span>,<\/span> 2<\/span>)<\/span><\/p>\n

# \u5206\u5272\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6982\u7387<\/span>
\n box,<\/span> cls =<\/span> x_cat.<\/span>split(<\/span>(<\/span>self.<\/span>reg_max *<\/span> 4<\/span>,<\/span> self.<\/span>nc)<\/span>,<\/span> 1<\/span>)<\/span>
\n dbox =<\/span> dist2bbox(<\/span>self.<\/span>dfl(<\/span>box)<\/span>,<\/span> self.<\/span>anchors.<\/span>unsqueeze(<\/span>0<\/span>)<\/span>,<\/span> xywh=<\/span>True<\/span>,<\/span> dim=<\/span>1<\/span>)<\/span> *<\/span> self.<\/span>strides<\/p>\n

# \u5f52\u4e00\u5316\u8fb9\u754c\u6846\u4ee5\u51cf\u5c0f\u91cf\u5316\u8bef\u5dee<\/span>
\n img_h =<\/span> shape[<\/span>2<\/span>]<\/span> *<\/span> self.<\/span>stride[<\/span>0<\/span>]<\/span>
\n img_w =<\/span> shape[<\/span>3<\/span>]<\/span> *<\/span> self.<\/span>stride[<\/span>0<\/span>]<\/span>
\n img_size =<\/span> torch.<\/span>tensor(<\/span>[<\/span>img_w,<\/span> img_h,<\/span> img_w,<\/span> img_h]<\/span>,<\/span> device=<\/span>dbox.<\/span>device)<\/span>.<\/span>reshape(<\/span>1<\/span>,<\/span> 4<\/span>,<\/span> 1<\/span>)<\/span>
\n dbox \/=<\/span> img_size<\/p>\n

# \u8fd4\u56de\u6700\u7ec8\u7684\u9884\u6d4b\u7ed3\u679c<\/span>
\n return<\/span> torch.<\/span>cat(<\/span>(<\/span>dbox,<\/span> cls.<\/span>sigmoid(<\/span>)<\/span>)<\/span>,<\/span> 1<\/span>)<\/span><\/p>\n

class<\/span> Segment<\/span>(<\/span>Detect)<\/span>:<\/span>
\n """YOLOv8\u5206\u5272\u5934，\u7528\u4e8e\u5206\u5272\u6a21\u578b\u3002"""<\/span><\/p>\n

def<\/span> __init__<\/span>(<\/span>self,<\/span> nc=<\/span>80<\/span>,<\/span> nm=<\/span>32<\/span>,<\/span> npr=<\/span>256<\/span>,<\/span> ch=<\/span>(<\/span>)<\/span>)<\/span>:<\/span>
\n """\u521d\u59cb\u5316\u5206\u5272\u6a21\u578b\u7684\u5c5e\u6027，\u5982\u63a9\u819c\u6570\u91cf\u3001\u539f\u578b\u6570\u91cf\u548c\u5377\u79ef\u5c42\u3002<\/p>\n

Args:
\n nc (int): \u7c7b\u522b\u6570，\u9ed8\u8ba4\u4e3a80\u3002
\n nm (int): \u63a9\u819c\u6570\u91cf，\u9ed8\u8ba4\u4e3a32\u3002
\n npr (int): \u539f\u578b\u6570\u91cf，\u9ed8\u8ba4\u4e3a256\u3002
\n ch (tuple): \u8f93\u5165\u901a\u9053\u6570\u7684\u5143\u7ec4\u3002
\n """<\/span>
\n super<\/span>(<\/span>)<\/span>.<\/span>__init__(<\/span>nc,<\/span> ch)<\/span>
\n self.<\/span>nm =<\/span> nm # \u63a9\u819c\u6570\u91cf<\/span>
\n self.<\/span>npr =<\/span> npr # \u539f\u578b\u6570\u91cf<\/span>
\n self.<\/span>proto =<\/span> Proto(<\/span>ch[<\/span>0<\/span>]<\/span>,<\/span> self.<\/span>npr,<\/span> self.<\/span>nm)<\/span> # \u539f\u578b<\/span>
\n self.<\/span>detect =<\/span> Detect.<\/span>forward<\/p>\n

c4 =<\/span> max<\/span>(<\/span>ch[<\/span>0<\/span>]<\/span> \/\/<\/span> 4<\/span>,<\/span> self.<\/span>nm)<\/span>
\n self.<\/span>cv4 =<\/span> nn.<\/span>ModuleList(<\/span>nn.<\/span>Sequential(<\/span>Conv(<\/span>x,<\/span> c4,<\/span> 3<\/span>)<\/span>,<\/span> Conv(<\/span>c4,<\/span> c4,<\/span> 3<\/span>)<\/span>,<\/span> nn.<\/span>Conv2d(<\/span>c4,<\/span> self.<\/span>nm,<\/span> 1<\/span>)<\/span>)<\/span> for<\/span> x in<\/span> ch)<\/span><\/p>\n

def<\/span> forward<\/span>(<\/span>self,<\/span> x)<\/span>:<\/span>
\n """\u8fd4\u56de\u6a21\u578b\u8f93\u51fa\u548c\u63a9\u819c\u7cfb\u6570，\u5982\u679c\u5728\u8bad\u7ec3\u6a21\u5f0f\u4e0b\u5219\u8fd4\u56de\u8f93\u51fa\u548c\u63a9\u819c\u7cfb\u6570\u3002<\/p>\n

Args:
\n x (list): \u8f93\u5165\u7279\u5f81\u56fe\u7684\u5217\u8868\u3002<\/p>\n

Returns:
\n Tuple: \u8f93\u51fa\u548c\u63a9\u819c\u7cfb\u6570\u3002
\n """<\/span>
\n p =<\/span> self.<\/span>proto(<\/span>x[<\/span>0<\/span>]<\/span>)<\/span> # \u63a9\u819c\u539f\u578b<\/span>
\n bs =<\/span> p.<\/span>shape[<\/span>0<\/span>]<\/span> # \u6279\u5927\u5c0f<\/span><\/p>\n

mc =<\/span> torch.<\/span>cat(<\/span>[<\/span>self.<\/span>cv4[<\/span>i]<\/span>(<\/span>x[<\/span>i]<\/span>)<\/span>.<\/span>view(<\/span>bs,<\/span> self.<\/span>nm,<\/span> –<\/span>1<\/span>)<\/span> for<\/span> i in<\/span> range<\/span>(<\/span>self.<\/span>nl)<\/span>]<\/span>,<\/span> 2<\/span>)<\/span> # \u63a9\u819c\u7cfb\u6570<\/span>
\n x =<\/span> self.<\/span>detect(<\/span>self,<\/span> x)<\/span><\/p>\n

if<\/span> self.<\/span>training:<\/span>
\n return<\/span> x,<\/span> mc,<\/span> p
\n return<\/span> (<\/span>torch.<\/span>cat(<\/span>[<\/span>x,<\/span> mc]<\/span>,<\/span> 1<\/span>)<\/span>,<\/span> p)<\/span> if<\/span> self.<\/span>export else<\/span> (<\/span>torch.<\/span>cat(<\/span>[<\/span>x[<\/span>0<\/span>]<\/span>,<\/span> mc]<\/span>,<\/span> 1<\/span>)<\/span>,<\/span> (<\/span>x[<\/span>1<\/span>]<\/span>,<\/span> mc,<\/span> p)<\/span>)<\/span><\/p>\n

class<\/span> Classify<\/span>(<\/span>nn.<\/span>Module)<\/span>:<\/span>
\n """YOLOv8\u5206\u7c7b\u5934，\u5c06\u8f93\u5165\u4ece (b, c1, 20, 20) \u8f6c\u6362\u4e3a (b, c2)\u3002"""<\/span><\/p>\n

def<\/span> __init__<\/span>(<\/span>self,<\/span> c1,<\/span> c2,<\/span> k=<\/span>1<\/span>,<\/span> s=<\/span>1<\/span>,<\/span> p=<\/span>None<\/span>,<\/span> g=<\/span>1<\/span>)<\/span>:<\/span>
\n """\u521d\u59cb\u5316YOLOv8\u5206\u7c7b\u5934，\u6307\u5b9a\u8f93\u5165\u548c\u8f93\u51fa\u901a\u9053\u3001\u5377\u79ef\u6838\u5927\u5c0f\u3001\u6b65\u5e45\u3001\u586b\u5145\u548c\u5206\u7ec4\u3002<\/p>\n

Args:
\n c1 (int): \u8f93\u5165\u901a\u9053\u6570\u3002
\n c2 (int): \u8f93\u51fa\u901a\u9053\u6570\u3002
\n k (int): \u5377\u79ef\u6838\u5927\u5c0f，\u9ed8\u8ba4\u4e3a1\u3002
\n s (int): \u6b65\u5e45，\u9ed8\u8ba4\u4e3a1\u3002
\n p (int): \u586b\u5145，\u9ed8\u8ba4\u4e3aNone\u3002
\n g (int): \u5206\u7ec4\u6570，\u9ed8\u8ba4\u4e3a1\u3002
\n """<\/span>
\n super<\/span>(<\/span>)<\/span>.<\/span>__init__(<\/span>)<\/span>
\n c_ =<\/span> 1280<\/span> # efficientnet_b0\u5927\u5c0f<\/span>
\n self.<\/span>conv =<\/span> Conv(<\/span>c1,<\/span> c_,<\/span> k,<\/span> s,<\/span> p,<\/span> g)<\/span>
\n self.<\/span>pool =<\/span> nn.<\/span>AdaptiveAvgPool2d(<\/span>1<\/span>)<\/span> # \u53d8\u6362\u4e3a (b, c_, 1, 1)<\/span>
\n self.<\/span>drop =<\/span> nn.<\/span>Dropout(<\/span>p=<\/span>0.0<\/span>,<\/span> inplace=<\/span>True<\/span>)<\/span>
\n self.<\/span>linear =<\/span> nn.<\/span>Linear(<\/span>c_,<\/span> c2)<\/span> # \u8f6c\u6362\u4e3a (b, c2)<\/span><\/p>\n

def<\/span> forward<\/span>(<\/span>self,<\/span> x)<\/span>:<\/span>
\n """\u5bf9\u8f93\u5165\u56fe\u50cf\u6570\u636e\u6267\u884c\u524d\u5411\u4f20\u64ad\u3002<\/p>\n

Args:
\n x (Tensor): \u8f93\u5165\u7279\u5f81\u56fe\u3002<\/p>\n

Returns:
\n Tensor: \u5206\u7c7b\u7ed3\u679c\u3002
\n """<\/span>
\n if<\/span> isinstance<\/span>(<\/span>x,<\/span> list<\/span>)<\/span>:<\/span>
\n x =<\/span> torch.<\/span>cat(<\/span>x,<\/span> 1<\/span>)<\/span>
\n x =<\/span> self.<\/span>linear(<\/span>self.<\/span>drop(<\/span>self.<\/span>pool(<\/span>self.<\/span>conv(<\/span>x)<\/span>)<\/span>.<\/span>flatten(<\/span>1<\/span>)<\/span>)<\/span>)<\/span>
\n return<\/span> x if<\/span> self.<\/span>training else<\/span> x.<\/span>softmax(<\/span>1<\/span>)<\/span><\/p>\n

# \u5176\u4ed6\u7c7b\u5982Pose\u548cRTDETRDecoder\u53ef\u4ee5\u6839\u636e\u9700\u8981\u6dfb\u52a0，\u7701\u7565\u4e86\u76f8\u4f3c\u7684\u7ed3\u6784\u548c\u6ce8\u91ca\u3002<\/span><\/p>\n

\u4ee3\u7801\u5206\u6790\u603b\u7ed3<\/h4>\n
  • Detect\u7c7b: \u4e3b\u8981\u8d1f\u8d23\u76ee\u6807\u68c0\u6d4b\u7684\u5b9e\u73b0，\u5305\u542b\u591a\u4e2a\u5377\u79ef\u5c42\u548c\u524d\u5411\u4f20\u64ad\u903b\u8f91\u3002\u5b83\u5904\u7406\u8f93\u5165\u7279\u5f81\u56fe\u5e76\u751f\u6210\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6982\u7387\u3002<\/li>\n
  • Segment\u7c7b: \u7ee7\u627f\u81eaDetect\u7c7b，\u6dfb\u52a0\u4e86\u5206\u5272\u76f8\u5173\u7684\u529f\u80fd，\u5305\u62ec\u63a9\u819c\u539f\u578b\u548c\u63a9\u819c\u7cfb\u6570\u7684\u751f\u6210\u3002<\/li>\n
  • Classify\u7c7b: \u5b9e\u73b0\u4e86\u5206\u7c7b\u529f\u80fd，\u5c06\u8f93\u5165\u7279\u5f81\u56fe\u8f6c\u6362\u4e3a\u7c7b\u522b\u9884\u6d4b\u3002<\/li>\n
  • \u6ce8\u91ca: \u6bcf\u4e2a\u7c7b\u548c\u65b9\u6cd5\u90fd\u5305\u542b\u8be6\u7ec6\u7684\u4e2d\u6587\u6ce8\u91ca，\u8bf4\u660e\u5176\u529f\u80fd\u3001\u53c2\u6570\u548c\u8fd4\u56de\u503c\u3002<\/li>\n

    \u4ee5\u4e0a\u662f\u5bf9\u4ee3\u7801\u7684\u6838\u5fc3\u90e8\u5206\u8fdb\u884c\u4e86\u7b80\u5316\u548c\u6ce8\u91ca，\u4fdd\u7559\u4e86\u4e3b\u8981\u7684\u529f\u80fd\u548c\u7ed3\u6784\u3002```
    \n\u8be5\u6587\u4ef6\u662fUltralytics YOLOv8\u6a21\u578b\u7684\u5934\u90e8\u6a21\u5757，\u4e3b\u8981\u5305\u542b\u4e86\u7528\u4e8e\u76ee\u6807\u68c0\u6d4b\u3001\u5206\u5272\u3001\u59ff\u6001\u4f30\u8ba1\u548c\u5206\u7c7b\u7684\u4e0d\u540c\u7c7b\u3002\u6587\u4ef6\u7684\u6838\u5fc3\u662f\u5b9a\u4e49\u4e86\u591a\u4e2a\u7c7b，\u6bcf\u4e2a\u7c7b\u5bf9\u5e94\u4e8e\u4e0d\u540c\u7684\u4efb\u52a1，\u5177\u4f53\u5982\u4e0b：<\/p>\n

    \u9996\u5148，Detect\u7c7b\u662fYOLOv8\u7684\u68c0\u6d4b\u5934，\u8d1f\u8d23\u5904\u7406\u76ee\u6807\u68c0\u6d4b\u4efb\u52a1\u3002\u5b83\u7684\u6784\u9020\u51fd\u6570\u63a5\u6536\u7c7b\u522b\u6570\u91cf\u548c\u901a\u9053\u6570，\u5e76\u521d\u59cb\u5316\u591a\u4e2a\u5377\u79ef\u5c42\u7528\u4e8e\u7279\u5f81\u63d0\u53d6\u3002\u8be5\u7c7b\u7684forward\u65b9\u6cd5\u63a5\u6536\u8f93\u5165\u7279\u5f81\u56fe，\u7ecf\u8fc7\u5377\u79ef\u5c42\u5904\u7406\u540e，\u751f\u6210\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6982\u7387\u3002\u5b83\u8fd8\u5305\u542b\u52a8\u6001\u951a\u6846\u751f\u6210\u548c\u8f93\u51fa\u683c\u5f0f\u5316\u7684\u903b\u8f91\u3002<\/p>\n

    \u63a5\u4e0b\u6765，Segment\u7c7b\u7ee7\u627f\u81eaDetect，\u7528\u4e8e\u56fe\u50cf\u5206\u5272\u4efb\u52a1\u3002\u5b83\u5728\u6784\u9020\u51fd\u6570\u4e2d\u589e\u52a0\u4e86\u63a9\u7801\u548c\u539f\u578b\u7684\u6570\u91cf，\u5e76\u5b9a\u4e49\u4e86\u989d\u5916\u7684\u5377\u79ef\u5c42\u7528\u4e8e\u751f\u6210\u63a9\u7801\u7cfb\u6570\u3002\u5728forward\u65b9\u6cd5\u4e2d，\u5b83\u4e0d\u4ec5\u8fd4\u56de\u68c0\u6d4b\u7ed3\u679c，\u8fd8\u8fd4\u56de\u63a9\u7801\u7cfb\u6570\u548c\u539f\u578b\u3002<\/p>\n

    Pose\u7c7b\u540c\u6837\u7ee7\u627f\u81eaDetect，\u7528\u4e8e\u59ff\u6001\u4f30\u8ba1\u3002\u5b83\u5b9a\u4e49\u4e86\u5173\u952e\u70b9\u7684\u5f62\u72b6，\u5e76\u5728forward\u65b9\u6cd5\u4e2d\u5904\u7406\u5173\u952e\u70b9\u7684\u9884\u6d4b，\u8fd4\u56de\u68c0\u6d4b\u7ed3\u679c\u548c\u5173\u952e\u70b9\u5750\u6807\u3002<\/p>\n

    Classify\u7c7b\u662fYOLOv8\u7684\u5206\u7c7b\u5934，\u8d1f\u8d23\u5c06\u8f93\u5165\u7684\u7279\u5f81\u56fe\u8f6c\u6362\u4e3a\u7c7b\u522b\u9884\u6d4b\u3002\u5b83\u4f7f\u7528\u5377\u79ef\u5c42\u3001\u6c60\u5316\u5c42\u548c\u5168\u8fde\u63a5\u5c42\u8fdb\u884c\u7279\u5f81\u63d0\u53d6\u548c\u5206\u7c7b\u3002<\/p>\n

    \u6700\u540e，RTDETRDecoder\u7c7b\u5b9e\u73b0\u4e86\u5b9e\u65f6\u53ef\u53d8\u5f62Transformer\u89e3\u7801\u5668，\u7ed3\u5408\u4e86Transformer\u67b6\u6784\u548c\u53ef\u53d8\u5f62\u5377\u79ef，\u7528\u4e8e\u751f\u6210\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u6807\u7b7e\u3002\u5b83\u7684\u6784\u9020\u51fd\u6570\u63a5\u6536\u591a\u4e2a\u53c2\u6570，\u5b9a\u4e49\u4e86\u8f93\u5165\u6295\u5f71\u3001Transformer\u6a21\u5757\u3001\u53bb\u566a\u90e8\u5206\u548c\u89e3\u7801\u5668\u5934\u3002\u5728forward\u65b9\u6cd5\u4e2d，\u5b83\u5904\u7406\u8f93\u5165\u7279\u5f81\u5e76\u751f\u6210\u6700\u7ec8\u7684\u9884\u6d4b\u7ed3\u679c\u3002<\/p>\n

    \u6574\u4f53\u800c\u8a00，\u8be5\u6587\u4ef6\u5b9a\u4e49\u4e86YOLOv8\u6a21\u578b\u7684\u591a\u4e2a\u5173\u952e\u7ec4\u4ef6，\u91c7\u7528\u6a21\u5757\u5316\u8bbe\u8ba1，\u4f7f\u5f97\u4e0d\u540c\u4efb\u52a1\u7684\u5b9e\u73b0\u76f8\u5bf9\u72ec\u7acb，\u4fbf\u4e8e\u6269\u5c55\u548c\u7ef4\u62a4\u3002\u6bcf\u4e2a\u7c7b\u90fd\u5b9e\u73b0\u4e86\u7279\u5b9a\u7684\u529f\u80fd，\u7ed3\u5408\u4e86\u6df1\u5ea6\u5b66\u4e60\u4e2d\u7684\u5e38\u7528\u6280\u672f，\u5982\u5377\u79ef\u3001\u6c60\u5316\u3001Transformer\u7b49，\u5c55\u793a\u4e86\u73b0\u4ee3\u76ee\u6807\u68c0\u6d4b\u6a21\u578b\u7684\u590d\u6742\u6027\u548c\u7075\u6d3b\u6027\u3002<\/p>\n

    ```python
    \nimport sys
    \nimport subprocess<\/p>\n

    def run_script(script_path):
    \n """
    \n \u4f7f\u7528\u5f53\u524d Python \u73af\u5883\u8fd0\u884c\u6307\u5b9a\u7684\u811a\u672c\u3002<\/p>\n

    Args:
    \n script_path (str): \u8981\u8fd0\u884c\u7684\u811a\u672c\u8def\u5f84<\/p>\n

    Returns:
    \n None
    \n """
    \n # \u83b7\u53d6\u5f53\u524d Python \u89e3\u91ca\u5668\u7684\u8def\u5f84
    \n python_path = sys.executable<\/p>\n

    # \u6784\u5efa\u8fd0\u884c\u547d\u4ee4，\u4f7f\u7528 streamlit \u8fd0\u884c\u6307\u5b9a\u7684\u811a\u672c
    \n command = f'"{python_path}" -m streamlit run "{script_path}"'<\/p>\n

    # \u6267\u884c\u547d\u4ee4，\u5e76\u7b49\u5f85\u5176\u5b8c\u6210
    \n result = subprocess.run(command, shell=True)<\/p>\n

    # \u68c0\u67e5\u547d\u4ee4\u6267\u884c\u7ed3\u679c，\u5982\u679c\u8fd4\u56de\u7801\u4e0d\u4e3a0，\u5219\u8868\u793a\u51fa\u9519
    \n if result.returncode != 0:
    \n print("\u811a\u672c\u8fd0\u884c\u51fa\u9519\u3002")<\/p>\n

    # \u4e3b\u7a0b\u5e8f\u5165\u53e3
    \nif __name__ == "__main__":
    \n # \u6307\u5b9a\u8981\u8fd0\u884c\u7684\u811a\u672c\u8def\u5f84
    \n script_path = "web.py" # \u5047\u8bbe\u811a\u672c\u5728\u5f53\u524d\u76ee\u5f55\u4e0b<\/p>\n

    # \u8c03\u7528\u51fd\u6570\u8fd0\u884c\u6307\u5b9a\u7684\u811a\u672c
    \n run_script(script_path)<\/p>\n

    \u4ee3\u7801\u6ce8\u91ca\u8bf4\u660e：<\/h4>\n
  • \n

    \u5bfc\u5165\u6a21\u5757：<\/p>\n

      \n
    • sys：\u7528\u4e8e\u83b7\u53d6\u5f53\u524d Python \u89e3\u91ca\u5668\u7684\u8def\u5f84\u3002<\/li>\n
    • subprocess：\u7528\u4e8e\u6267\u884c\u5916\u90e8\u547d\u4ee4\u3002<\/li>\n<\/ul>\n<\/li>\n
    • \n

      run_script \u51fd\u6570：<\/p>\n

        \n
      • \u8be5\u51fd\u6570\u63a5\u6536\u4e00\u4e2a\u811a\u672c\u8def\u5f84\u4f5c\u4e3a\u53c2\u6570，\u5e76\u4f7f\u7528\u5f53\u524d Python \u73af\u5883\u8fd0\u884c\u8be5\u811a\u672c\u3002<\/li>\n
      • python_path = sys.executable：\u83b7\u53d6\u5f53\u524d Python \u89e3\u91ca\u5668\u7684\u5b8c\u6574\u8def\u5f84\u3002<\/li>\n
      • command：\u6784\u5efa\u8981\u6267\u884c\u7684\u547d\u4ee4\u5b57\u7b26\u4e32，\u4f7f\u7528 streamlit \u6a21\u5757\u8fd0\u884c\u6307\u5b9a\u7684\u811a\u672c\u3002<\/li>\n
      • subprocess.run(command, shell=True)：\u6267\u884c\u6784\u5efa\u7684\u547d\u4ee4，\u5e76\u7b49\u5f85\u5176\u5b8c\u6210\u3002<\/li>\n
      • result.returncode：\u68c0\u67e5\u547d\u4ee4\u7684\u8fd4\u56de\u7801，\u82e5\u4e0d\u4e3a0，\u5219\u8868\u793a\u6267\u884c\u8fc7\u7a0b\u4e2d\u51fa\u73b0\u9519\u8bef\u3002<\/li>\n<\/ul>\n<\/li>\n
      • \n

        \u4e3b\u7a0b\u5e8f\u5165\u53e3：<\/p>\n

          \n
        • \u4f7f\u7528 if __name__ == "__main__": \u786e\u4fdd\u53ea\u6709\u5728\u76f4\u63a5\u8fd0\u884c\u8be5\u811a\u672c\u65f6\u624d\u4f1a\u6267\u884c\u4ee5\u4e0b\u4ee3\u7801\u3002<\/li>\n
        • script_path：\u6307\u5b9a\u8981\u8fd0\u884c\u7684\u811a\u672c\u8def\u5f84（\u5728\u6b64\u793a\u4f8b\u4e2d\u4e3a "web.py"）\u3002<\/li>\n
        • \u8c03\u7528 run_script(script_path) \u6765\u6267\u884c\u6307\u5b9a\u7684\u811a\u672c\u3002```
          \n\u8fd9\u4e2a\u7a0b\u5e8f\u6587\u4ef6\u540d\u4e3a ui.py，\u4e3b\u8981\u529f\u80fd\u662f\u4f7f\u7528\u5f53\u524d\u7684 Python \u73af\u5883\u6765\u8fd0\u884c\u4e00\u4e2a\u6307\u5b9a\u7684\u811a\u672c\u3002\u7a0b\u5e8f\u9996\u5148\u5bfc\u5165\u4e86\u5fc5\u8981\u7684\u6a21\u5757，\u5305\u62ec sys\u3001os \u548c subprocess，\u4ee5\u53ca\u4e00\u4e2a\u81ea\u5b9a\u4e49\u6a21\u5757 QtFusion.path \u4e2d\u7684 abs_path \u51fd\u6570\u3002<\/li>\n<\/ul>\n<\/li>\n

          \u5728 run_script \u51fd\u6570\u4e2d，\u7a0b\u5e8f\u63a5\u6536\u4e00\u4e2a\u53c2\u6570 script_path，\u8fd9\u4e2a\u53c2\u6570\u662f\u8981\u8fd0\u884c\u7684\u811a\u672c\u7684\u8def\u5f84\u3002\u51fd\u6570\u5185\u90e8\u9996\u5148\u83b7\u53d6\u5f53\u524d Python \u89e3\u91ca\u5668\u7684\u8def\u5f84，\u5e76\u5c06\u5176\u5b58\u50a8\u5728 python_path \u53d8\u91cf\u4e2d\u3002\u63a5\u7740，\u7a0b\u5e8f\u6784\u5efa\u4e86\u4e00\u4e2a\u547d\u4ee4\u5b57\u7b26\u4e32，\u8fd9\u4e2a\u547d\u4ee4\u4f7f\u7528 streamlit \u6765\u8fd0\u884c\u6307\u5b9a\u7684\u811a\u672c\u3002\u547d\u4ee4\u7684\u683c\u5f0f\u662f "{python_path}" -m streamlit run "{script_path}"，\u5176\u4e2d {python_path} \u548c {script_path} \u4f1a\u88ab\u5b9e\u9645\u7684\u8def\u5f84\u66ff\u6362\u3002<\/p>\n

          \u968f\u540e，\u7a0b\u5e8f\u4f7f\u7528 subprocess.run \u65b9\u6cd5\u6765\u6267\u884c\u8fd9\u4e2a\u547d\u4ee4\u3002shell=True \u53c2\u6570\u5141\u8bb8\u5728 shell \u4e2d\u6267\u884c\u547d\u4ee4\u3002\u6267\u884c\u540e，\u7a0b\u5e8f\u68c0\u67e5\u8fd4\u56de\u7684\u7ed3\u679c，\u5982\u679c\u8fd4\u56de\u7801\u4e0d\u4e3a 0，\u8868\u793a\u811a\u672c\u8fd0\u884c\u51fa\u9519，\u7a0b\u5e8f\u4f1a\u6253\u5370\u51fa\u201c\u811a\u672c\u8fd0\u884c\u51fa\u9519\u3002\u201d\u7684\u63d0\u793a\u4fe1\u606f\u3002<\/p>\n

          \u5728\u6587\u4ef6\u7684\u6700\u540e\u90e8\u5206，\u7a0b\u5e8f\u901a\u8fc7 if __name__ == "__main__": \u5224\u65ad\u662f\u5426\u662f\u76f4\u63a5\u8fd0\u884c\u8be5\u811a\u672c\u3002\u5982\u679c\u662f，\u7a0b\u5e8f\u4f1a\u8c03\u7528 abs_path \u51fd\u6570\u6765\u83b7\u53d6 web.py \u811a\u672c\u7684\u7edd\u5bf9\u8def\u5f84，\u5e76\u5c06\u5176\u4f20\u9012\u7ed9 run_script \u51fd\u6570\u4ee5\u6267\u884c\u8be5\u811a\u672c\u3002<\/p>\n

          \u603b\u7684\u6765\u8bf4，\u8fd9\u4e2a\u7a0b\u5e8f\u7684\u4e3b\u8981\u76ee\u7684\u662f\u4e3a\u4e86\u65b9\u4fbf\u5730\u8fd0\u884c\u4e00\u4e2a\u540d\u4e3a web.py \u7684\u811a\u672c，\u786e\u4fdd\u4f7f\u7528\u7684\u662f\u5f53\u524d\u7684 Python \u73af\u5883，\u5e76\u80fd\u591f\u5904\u7406\u53ef\u80fd\u51fa\u73b0\u7684\u9519\u8bef\u3002<\/p>\n

          ```python
          \nimport torch
          \nfrom ultralytics.engine.predictor import BasePredictor
          \nfrom ultralytics.engine.results import Results
          \nfrom ultralytics.utils import ops<\/p>\n

          class NASPredictor(BasePredictor):
          \n """
          \n Ultralytics YOLO NAS \u9884\u6d4b\u5668，\u7528\u4e8e\u76ee\u6807\u68c0\u6d4b\u3002<\/p>\n

          \u8be5\u7c7b\u6269\u5c55\u4e86 Ultralytics \u5f15\u64ce\u4e2d\u7684 `BasePredictor`，\u8d1f\u8d23\u5bf9 YOLO NAS \u6a21\u578b\u751f\u6210\u7684\u539f\u59cb\u9884\u6d4b\u7ed3\u679c\u8fdb\u884c\u540e\u5904\u7406\u3002
          \n \u5b83\u5e94\u7528\u4e86\u975e\u6781\u5927\u503c\u6291\u5236（NMS）\u548c\u5c06\u8fb9\u754c\u6846\u7f29\u653e\u5230\u539f\u59cb\u56fe\u50cf\u5c3a\u5bf8\u7b49\u64cd\u4f5c\u3002<\/p>\n

          \u5c5e\u6027:
          \n args (Namespace): \u5305\u542b\u5404\u79cd\u540e\u5904\u7406\u914d\u7f6e\u7684\u547d\u540d\u7a7a\u95f4\u3002
          \n """<\/p>\n

          def postprocess(self, preds_in, img, orig_imgs):
          \n """\u5bf9\u9884\u6d4b\u7ed3\u679c\u8fdb\u884c\u540e\u5904\u7406，\u5e76\u8fd4\u56de Results \u5bf9\u8c61\u7684\u5217\u8868\u3002"""<\/p>\n

          # \u5c06\u9884\u6d4b\u6846\u4ece xyxy \u683c\u5f0f\u8f6c\u6362\u4e3a xywh \u683c\u5f0f，\u5e76\u63d0\u53d6\u7c7b\u5206\u6570
          \n boxes = ops.xyxy2xywh(preds_in[0][0]) # \u83b7\u53d6\u8fb9\u754c\u6846
          \n preds = torch.cat((boxes, preds_in[0][1]), -1).permute(0, 2, 1) # \u5408\u5e76\u8fb9\u754c\u6846\u548c\u7c7b\u5206\u6570<\/p>\n

          # \u5e94\u7528\u975e\u6781\u5927\u503c\u6291\u5236，\u8fc7\u6ee4\u91cd\u53e0\u7684\u8fb9\u754c\u6846
          \n preds = ops.non_max_suppression(preds,
          \n self.args.conf, # \u7f6e\u4fe1\u5ea6\u9608\u503c
          \n self.args.iou, # IOU \u9608\u503c
          \n agnostic=self.args.agnostic_nms, # \u662f\u5426\u5bf9\u7c7b\u522b\u4e0d\u654f\u611f
          \n max_det=self.args.max_det, # \u6700\u5927\u68c0\u6d4b\u6570\u91cf
          \n classes=self.args.classes) # \u9009\u62e9\u7684\u7c7b\u522b<\/p>\n

          # \u5982\u679c\u8f93\u5165\u56fe\u50cf\u4e0d\u662f\u5217\u8868，\u5219\u5c06\u5176\u8f6c\u6362\u4e3a numpy \u6570\u7ec4
          \n if not isinstance(orig_imgs, list):
          \n orig_imgs = ops.convert_torch2numpy_batch(orig_imgs)<\/p>\n

          results = [] # \u5b58\u50a8\u7ed3\u679c\u7684\u5217\u8868
          \n for i, pred in enumerate(preds):
          \n orig_img = orig_imgs[i] # \u83b7\u53d6\u539f\u59cb\u56fe\u50cf
          \n # \u5c06\u9884\u6d4b\u6846\u7f29\u653e\u5230\u539f\u59cb\u56fe\u50cf\u7684\u5c3a\u5bf8
          \n pred[:, :4] = ops.scale_boxes(img.shape[2:], pred[:, :4], orig_img.shape)
          \n img_path = self.batch[0][i] # \u83b7\u53d6\u56fe\u50cf\u8def\u5f84
          \n # \u521b\u5efa Results \u5bf9\u8c61\u5e76\u6dfb\u52a0\u5230\u7ed3\u679c\u5217\u8868
          \n results.append(Results(orig_img, path=img_path, names=self.model.names, boxes=pred))<\/p>\n

          return results # \u8fd4\u56de\u5904\u7406\u540e\u7684\u7ed3\u679c\u5217\u8868<\/p>\n

          \u4ee3\u7801\u6838\u5fc3\u90e8\u5206\u8bf4\u660e：<\/h4>\n
        • \u7c7b\u5b9a\u4e49：NASPredictor \u7ee7\u627f\u81ea BasePredictor，\u7528\u4e8e\u5904\u7406 YOLO NAS \u6a21\u578b\u7684\u9884\u6d4b\u7ed3\u679c\u3002<\/li>\n
        • postprocess \u65b9\u6cd5：\u8be5\u65b9\u6cd5\u662f\u540e\u5904\u7406\u7684\u6838\u5fc3，\u8d1f\u8d23\u5c06\u6a21\u578b\u7684\u539f\u59cb\u9884\u6d4b\u7ed3\u679c\u8f6c\u6362\u4e3a\u53ef\u7528\u7684\u683c\u5f0f\u3002\n
            \n
          • \u8fb9\u754c\u6846\u8f6c\u6362：\u5c06\u9884\u6d4b\u7684\u8fb9\u754c\u6846\u4ece xyxy \u683c\u5f0f\u8f6c\u6362\u4e3a xywh \u683c\u5f0f，\u5e76\u5408\u5e76\u7c7b\u5206\u6570\u3002<\/li>\n
          • \u975e\u6781\u5927\u503c\u6291\u5236：\u901a\u8fc7 NMS \u8fc7\u6ee4\u6389\u91cd\u53e0\u7684\u8fb9\u754c\u6846，\u4ee5\u4fdd\u7559\u6700\u4f18\u7684\u68c0\u6d4b\u7ed3\u679c\u3002<\/li>\n
          • \u56fe\u50cf\u5904\u7406：\u786e\u4fdd\u539f\u59cb\u56fe\u50cf\u683c\u5f0f\u6b63\u786e，\u5e76\u5c06\u9884\u6d4b\u6846\u7f29\u653e\u5230\u539f\u59cb\u56fe\u50cf\u7684\u5c3a\u5bf8\u3002<\/li>\n
          • \u7ed3\u679c\u6536\u96c6：\u5c06\u6bcf\u4e2a\u56fe\u50cf\u7684\u9884\u6d4b\u7ed3\u679c\u5c01\u88c5\u6210 Results \u5bf9\u8c61\u5e76\u8fd4\u56de\u3002```
            \n\u8fd9\u4e2a\u7a0b\u5e8f\u6587\u4ef6\u662fUltralytics YOLO NAS\u6a21\u578b\u7684\u9884\u6d4b\u5668\u5b9e\u73b0，\u4e3b\u8981\u7528\u4e8e\u76ee\u6807\u68c0\u6d4b\u4efb\u52a1\u3002\u5b83\u7ee7\u627f\u81eaUltralytics\u5f15\u64ce\u4e2d\u7684BasePredictor\u7c7b，\u8d1f\u8d23\u5bf9YOLO NAS\u6a21\u578b\u751f\u6210\u7684\u539f\u59cb\u9884\u6d4b\u7ed3\u679c\u8fdb\u884c\u540e\u5904\u7406\u3002\u540e\u5904\u7406\u7684\u4e3b\u8981\u64cd\u4f5c\u5305\u62ec\u975e\u6781\u5927\u503c\u6291\u5236（NMS）\u548c\u5c06\u8fb9\u754c\u6846\u7f29\u653e\u5230\u539f\u59cb\u56fe\u50cf\u7684\u5c3a\u5bf8\u3002<\/li>\n<\/ul>\n<\/li>\n

            \u5728\u7c7b\u7684\u6587\u6863\u5b57\u7b26\u4e32\u4e2d，\u9996\u5148\u4ecb\u7ecd\u4e86\u8fd9\u4e2a\u7c7b\u7684\u529f\u80fd\u548c\u7528\u9014，\u8bf4\u660e\u5b83\u4e3b\u8981\u7528\u4e8e\u5904\u7406YOLO NAS\u6a21\u578b\u7684\u8f93\u51fa，\u5e76\u4e14\u63d0\u4f9b\u4e86\u4e00\u4e2a\u793a\u4f8b\u4ee3\u7801，\u5c55\u793a\u5982\u4f55\u4f7f\u7528\u8fd9\u4e2a\u9884\u6d4b\u5668\u8fdb\u884c\u540e\u5904\u7406\u3002\u9700\u8981\u6ce8\u610f\u7684\u662f，\u8fd9\u4e2a\u7c7b\u901a\u5e38\u4e0d\u4f1a\u88ab\u76f4\u63a5\u5b9e\u4f8b\u5316，\u800c\u662f\u4f5c\u4e3aNAS\u7c7b\u5185\u90e8\u7684\u4e00\u4e2a\u7ec4\u4ef6\u4f7f\u7528\u3002<\/p>\n

            postprocess\u65b9\u6cd5\u662f\u8fd9\u4e2a\u7c7b\u7684\u6838\u5fc3\u529f\u80fd，\u63a5\u6536\u539f\u59cb\u9884\u6d4b\u7ed3\u679c\u3001\u8f93\u5165\u56fe\u50cf\u548c\u539f\u59cb\u56fe\u50cf\u5217\u8868\u4f5c\u4e3a\u53c2\u6570\u3002\u9996\u5148，\u5b83\u5c06\u539f\u59cb\u9884\u6d4b\u7ed3\u679c\u4e2d\u7684\u8fb9\u754c\u6846\u8f6c\u6362\u4e3aYOLO\u683c\u5f0f，\u5e76\u5c06\u8fb9\u754c\u6846\u548c\u7c7b\u522b\u5206\u6570\u5408\u5e76\u3002\u63a5\u7740，\u8c03\u7528\u975e\u6781\u5927\u503c\u6291\u5236\u51fd\u6570\u6765\u8fc7\u6ee4\u6389\u91cd\u53e0\u7684\u8fb9\u754c\u6846，\u4ee5\u63d0\u9ad8\u68c0\u6d4b\u7ed3\u679c\u7684\u51c6\u786e\u6027\u3002<\/p>\n

            \u5728\u5904\u7406\u539f\u59cb\u56fe\u50cf\u65f6，\u5982\u679c\u8f93\u5165\u7684\u56fe\u50cf\u4e0d\u662f\u5217\u8868\u5f62\u5f0f，\u800c\u662f\u4e00\u4e2aTorch\u5f20\u91cf，\u5219\u5c06\u5176\u8f6c\u6362\u4e3aNumPy\u6570\u7ec4\u683c\u5f0f\u3002\u968f\u540e，\u7a0b\u5e8f\u904d\u5386\u6bcf\u4e2a\u9884\u6d4b\u7ed3\u679c，\u4f7f\u7528scale_boxes\u51fd\u6570\u5c06\u8fb9\u754c\u6846\u7f29\u653e\u5230\u539f\u59cb\u56fe\u50cf\u7684\u5c3a\u5bf8，\u5e76\u521b\u5efa\u4e00\u4e2aResults\u5bf9\u8c61，\u5305\u542b\u539f\u59cb\u56fe\u50cf\u3001\u56fe\u50cf\u8def\u5f84\u3001\u7c7b\u522b\u540d\u79f0\u548c\u5904\u7406\u540e\u7684\u8fb9\u754c\u6846\u4fe1\u606f\u3002\u6700\u7ec8，\u8fd4\u56de\u4e00\u4e2a\u5305\u542b\u6240\u6709\u7ed3\u679c\u7684\u5217\u8868\u3002<\/p>\n

            \u6574\u4f53\u6765\u770b，\u8fd9\u4e2a\u6587\u4ef6\u7684\u4e3b\u8981\u529f\u80fd\u662f\u5c06YOLO NAS\u6a21\u578b\u7684\u9884\u6d4b\u7ed3\u679c\u8fdb\u884c\u540e\u5904\u7406，\u4ee5\u4fbf\u4e8e\u540e\u7eed\u7684\u7ed3\u679c\u5c55\u793a\u548c\u5206\u6790\u3002<\/p>\n

            ```python
            \nimport os
            \nimport torch
            \nimport yaml
            \nfrom ultralytics import YOLO # \u5bfc\u5165YOLO\u6a21\u578b<\/p>\n

            if __name__ == '__main__': # \u786e\u4fdd\u8be5\u6a21\u5757\u88ab\u76f4\u63a5\u8fd0\u884c\u65f6\u624d\u6267\u884c\u4ee5\u4e0b\u4ee3\u7801
            \n # \u8bbe\u7f6e\u8bad\u7ec3\u53c2\u6570
            \n workers = 1 # \u6570\u636e\u52a0\u8f7d\u7684\u5de5\u4f5c\u8fdb\u7a0b\u6570
            \n batch = 8 # \u6bcf\u4e2a\u6279\u6b21\u7684\u6837\u672c\u6570\u91cf，\u9700\u6839\u636e\u663e\u5b58\u548c\u5185\u5b58\u8fdb\u884c\u8c03\u6574
            \n device = "0" if torch.cuda.is_available() else "cpu" # \u5224\u65ad\u662f\u5426\u4f7f\u7528GPU<\/p>\n

            # \u83b7\u53d6\u6570\u636e\u96c6\u7684yaml\u914d\u7f6e\u6587\u4ef6\u7684\u7edd\u5bf9\u8def\u5f84
            \n data_path = abs_path(f'datasets\/data\/data.yaml', path_type='current') <\/p>\n

            # \u8bfb\u53d6YAML\u6587\u4ef6，\u4fdd\u6301\u539f\u6709\u987a\u5e8f
            \n with open(data_path, 'r') as file:
            \n data = yaml.load(file, Loader=yaml.FullLoader)<\/p>\n

            # \u4fee\u6539\u6570\u636e\u96c6\u8def\u5f84
            \n if 'train' in data and 'val' in data and 'test' in data:
            \n directory_path = os.path.dirname(data_path.replace(os.sep, '\/')) # \u83b7\u53d6\u76ee\u5f55\u8def\u5f84
            \n data['train'] = directory_path + '\/train' # \u66f4\u65b0\u8bad\u7ec3\u96c6\u8def\u5f84
            \n data['val'] = directory_path + '\/val' # \u66f4\u65b0\u9a8c\u8bc1\u96c6\u8def\u5f84
            \n data['test'] = directory_path + '\/test' # \u66f4\u65b0\u6d4b\u8bd5\u96c6\u8def\u5f84<\/p>\n

            # \u5c06\u4fee\u6539\u540e\u7684\u6570\u636e\u5199\u56deYAML\u6587\u4ef6
            \n with open(data_path, 'w') as file:
            \n yaml.safe_dump(data, file, sort_keys=False)<\/p>\n

            # \u52a0\u8f7dYOLO\u6a21\u578b\u914d\u7f6e\u548c\u9884\u8bad\u7ec3\u6743\u91cd
            \n model = YOLO(r"C:\\\\codeseg\\\\codenew\\\\50+\u79cdYOLOv8\u7b97\u6cd5\u6539\u8fdb\u6e90\u7801\u5927\u5168\u548c\u8c03\u8bd5\u52a0\u8f7d\u8bad\u7ec3\u6559\u7a0b（\u975e\u5fc5\u8981）\\\\\u6539\u8fdbYOLOv8\u6a21\u578b\u914d\u7f6e\u6587\u4ef6\\\\yolov8-seg-C2f-Faster.yaml").load(".\/weights\/yolov8s-seg.pt")<\/p>\n

            # \u5f00\u59cb\u8bad\u7ec3\u6a21\u578b
            \n results = model.train(
            \n data=data_path, # \u6307\u5b9a\u8bad\u7ec3\u6570\u636e\u7684\u914d\u7f6e\u6587\u4ef6\u8def\u5f84
            \n device=device, # \u6307\u5b9a\u4f7f\u7528\u7684\u8bbe\u5907（GPU\u6216CPU）
            \n workers=workers, # \u6307\u5b9a\u6570\u636e\u52a0\u8f7d\u7684\u5de5\u4f5c\u8fdb\u7a0b\u6570
            \n imgsz=640, # \u8f93\u5165\u56fe\u50cf\u7684\u5927\u5c0f
            \n epochs=100, # \u8bad\u7ec3\u7684\u8f6e\u6570
            \n batch=batch, # \u6bcf\u4e2a\u6279\u6b21\u7684\u6837\u672c\u6570\u91cf
            \n )<\/p>\n

            \u4ee3\u7801\u6ce8\u91ca\u8bf4\u660e：<\/h4>\n
          • \u5bfc\u5165\u5fc5\u8981\u7684\u5e93：\u5bfc\u5165\u4e86\u5904\u7406\u6587\u4ef6\u8def\u5f84\u3001\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6（PyTorch）\u3001YAML\u6587\u4ef6\u5904\u7406\u548cYOLO\u6a21\u578b\u7684\u5e93\u3002<\/li>\n
          • \u8bbe\u7f6e\u8bad\u7ec3\u53c2\u6570：\u5b9a\u4e49\u4e86\u6570\u636e\u52a0\u8f7d\u7684\u5de5\u4f5c\u8fdb\u7a0b\u6570\u3001\u6279\u6b21\u5927\u5c0f\u548c\u8bbe\u5907\u7c7b\u578b（GPU\u6216CPU）\u3002<\/li>\n
          • \u83b7\u53d6\u6570\u636e\u96c6\u8def\u5f84：\u901a\u8fc7abs_path\u51fd\u6570\u83b7\u53d6\u6570\u636e\u96c6\u914d\u7f6e\u6587\u4ef6\u7684\u7edd\u5bf9\u8def\u5f84\u3002<\/li>\n
          • \u8bfb\u53d6\u548c\u4fee\u6539YAML\u6587\u4ef6：\u8bfb\u53d6YAML\u6587\u4ef6，\u66f4\u65b0\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5\u96c6\u7684\u8def\u5f84，\u5e76\u5c06\u4fee\u6539\u540e\u7684\u5185\u5bb9\u5199\u56de\u6587\u4ef6\u3002<\/li>\n
          • \u52a0\u8f7dYOLO\u6a21\u578b：\u4f7f\u7528\u6307\u5b9a\u7684\u914d\u7f6e\u6587\u4ef6\u548c\u9884\u8bad\u7ec3\u6743\u91cd\u52a0\u8f7dYOLO\u6a21\u578b\u3002<\/li>\n
          • \u8bad\u7ec3\u6a21\u578b：\u8c03\u7528train\u65b9\u6cd5\u5f00\u59cb\u8bad\u7ec3，\u4f20\u5165\u5fc5\u8981\u7684\u53c2\u6570，\u5982\u6570\u636e\u8def\u5f84\u3001\u8bbe\u5907\u3001\u5de5\u4f5c\u8fdb\u7a0b\u6570\u3001\u56fe\u50cf\u5927\u5c0f\u3001\u8bad\u7ec3\u8f6e\u6570\u548c\u6279\u6b21\u5927\u5c0f\u3002```
            \n\u8be5\u7a0b\u5e8f\u6587\u4ef6train.py\u4e3b\u8981\u7528\u4e8e\u8bad\u7ec3YOLO（You Only Look Once）\u6a21\u578b，\u5177\u4f53\u662fYOLOv8\u7684\u4e00\u4e2a\u53d8\u79cd\u3002\u7a0b\u5e8f\u9996\u5148\u5bfc\u5165\u4e86\u4e00\u4e9b\u5fc5\u8981\u7684\u5e93，\u5305\u62ec\u64cd\u4f5c\u7cfb\u7edf\u5e93os\u3001\u6df1\u5ea6\u5b66\u4e60\u6846\u67b6torch\u3001YAML\u5904\u7406\u5e93yaml\u4ee5\u53caYOLO\u6a21\u578b\u5e93ultralytics\u3002\u6b64\u5916，\u8fd8\u8bbe\u7f6e\u4e86Matplotlib\u7684\u540e\u7aef\u4e3a\u2019TkAgg\u2019，\u4ee5\u4fbf\u4e8e\u53ef\u89c6\u5316\u3002<\/li>\n

            \u5728\u4e3b\u7a0b\u5e8f\u4e2d，\u9996\u5148\u5b9a\u4e49\u4e86\u4e00\u4e9b\u8bad\u7ec3\u53c2\u6570，\u5305\u62ec\u5de5\u4f5c\u8fdb\u7a0b\u6570workers\u3001\u6279\u6b21\u5927\u5c0fbatch\u548c\u8bbe\u5907\u7c7b\u578bdevice\u3002\u8bbe\u5907\u7c7b\u578b\u4f1a\u6839\u636e\u5f53\u524d\u662f\u5426\u6709\u53ef\u7528\u7684GPU（CUDA）\u6765\u51b3\u5b9a，\u5982\u679c\u6709\u5219\u4f7f\u7528GPU，\u5426\u5219\u4f7f\u7528CPU\u3002\u63a5\u7740，\u7a0b\u5e8f\u901a\u8fc7abs_path\u51fd\u6570\u83b7\u53d6\u6570\u636e\u96c6\u914d\u7f6e\u6587\u4ef6\u7684\u7edd\u5bf9\u8def\u5f84，\u8fd9\u4e2a\u914d\u7f6e\u6587\u4ef6\u662f\u4e00\u4e2aYAML\u683c\u5f0f\u7684\u6587\u4ef6，\u91cc\u9762\u5305\u542b\u4e86\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5\u6570\u636e\u7684\u8def\u5f84\u3002<\/p>\n

            \u7a0b\u5e8f\u8bfb\u53d6\u8be5YAML\u6587\u4ef6，\u5e76\u68c0\u67e5\u5176\u4e2d\u662f\u5426\u5305\u542b\u2019train\u2019\u3001'val\u2019\u548c\u2019test\u2019\u8fd9\u51e0\u4e2a\u952e\u3002\u5982\u679c\u5b58\u5728，\u7a0b\u5e8f\u4f1a\u6839\u636e\u6570\u636e\u96c6\u7684\u76ee\u5f55\u8def\u5f84\u66f4\u65b0\u8fd9\u4e9b\u952e\u7684\u503c，\u786e\u4fdd\u5b83\u4eec\u6307\u5411\u6b63\u786e\u7684\u8bad\u7ec3\u3001\u9a8c\u8bc1\u548c\u6d4b\u8bd5\u6570\u636e\u76ee\u5f55\u3002\u66f4\u65b0\u5b8c\u6210\u540e，\u7a0b\u5e8f\u5c06\u4fee\u6539\u540e\u7684\u6570\u636e\u5199\u56de\u5230YAML\u6587\u4ef6\u4e2d\u3002<\/p>\n

            \u63a5\u4e0b\u6765，\u7a0b\u5e8f\u52a0\u8f7dYOLO\u6a21\u578b\u7684\u914d\u7f6e\u6587\u4ef6\u548c\u9884\u8bad\u7ec3\u6743\u91cd\u3002\u8fd9\u91cc\u7684\u6a21\u578b\u914d\u7f6e\u6587\u4ef6\u8def\u5f84\u662f\u786c\u7f16\u7801\u7684，\u7528\u6237\u53ef\u4ee5\u6839\u636e\u9700\u8981\u4fee\u6539\u3002\u6a21\u578b\u52a0\u8f7d\u5b8c\u6210\u540e，\u7a0b\u5e8f\u8c03\u7528model.train\u65b9\u6cd5\u5f00\u59cb\u8bad\u7ec3，\u4f20\u5165\u7684\u6570\u636e\u914d\u7f6e\u6587\u4ef6\u8def\u5f84\u3001\u8bbe\u5907\u7c7b\u578b\u3001\u5de5\u4f5c\u8fdb\u7a0b\u6570\u3001\u8f93\u5165\u56fe\u50cf\u5927\u5c0f\u3001\u8bad\u7ec3\u8f6e\u6570\u548c\u6279\u6b21\u5927\u5c0f\u7b49\u53c2\u6570\u3002<\/p>\n

            \u6574\u4f53\u6765\u770b，\u8be5\u7a0b\u5e8f\u7684\u529f\u80fd\u662f\u914d\u7f6e\u548c\u542f\u52a8YOLOv8\u6a21\u578b\u7684\u8bad\u7ec3\u8fc7\u7a0b，\u7528\u6237\u53ef\u4ee5\u6839\u636e\u81ea\u5df1\u7684\u786c\u4ef6\u6761\u4ef6\u548c\u6570\u636e\u96c6\u8fdb\u884c\u76f8\u5e94\u7684\u8c03\u6574\u3002\u7a0b\u5e8f\u7684\u7075\u6d3b\u6027\u4f53\u73b0\u5728\u53ef\u4ee5\u901a\u8fc7\u4fee\u6539YAML\u6587\u4ef6\u548c\u6a21\u578b\u914d\u7f6e\u6587\u4ef6\u6765\u9002\u5e94\u4e0d\u540c\u7684\u8bad\u7ec3\u9700\u6c42\u3002<\/p>\n

            ```python
            \nclass BaseDataset(Dataset):
            \n """
            \n \u57fa\u7840\u6570\u636e\u96c6\u7c7b，\u7528\u4e8e\u52a0\u8f7d\u548c\u5904\u7406\u56fe\u50cf\u6570\u636e\u3002<\/p>\n

            \u53c2\u6570:
            \n img_path (str): \u56fe\u50cf\u6587\u4ef6\u5939\u7684\u8def\u5f84\u3002
            \n imgsz (int, optional): \u56fe\u50cf\u5927\u5c0f，\u9ed8\u8ba4\u4e3a640\u3002
            \n cache (bool, optional): \u5728\u8bad\u7ec3\u671f\u95f4\u5c06\u56fe\u50cf\u7f13\u5b58\u5230RAM\u6216\u78c1\u76d8，\u9ed8\u8ba4\u4e3aFalse\u3002
            \n augment (bool, optional): \u5982\u679c\u4e3aTrue，\u5219\u5e94\u7528\u6570\u636e\u589e\u5f3a，\u9ed8\u8ba4\u4e3aTrue\u3002
            \n hyp (dict, optional): \u5e94\u7528\u6570\u636e\u589e\u5f3a\u7684\u8d85\u53c2\u6570，\u9ed8\u8ba4\u4e3aNone\u3002
            \n prefix (str, optional): \u65e5\u5fd7\u6d88\u606f\u4e2d\u6253\u5370\u7684\u524d\u7f00，\u9ed8\u8ba4\u4e3a''\u3002
            \n rect (bool, optional): \u5982\u679c\u4e3aTrue，\u5219\u4f7f\u7528\u77e9\u5f62\u8bad\u7ec3，\u9ed8\u8ba4\u4e3aFalse\u3002
            \n batch_size (int, optional): \u6279\u6b21\u5927\u5c0f，\u9ed8\u8ba4\u4e3aNone\u3002
            \n stride (int, optional): \u6b65\u5e45，\u9ed8\u8ba4\u4e3a32\u3002
            \n pad (float, optional): \u586b\u5145，\u9ed8\u8ba4\u4e3a0.0\u3002
            \n single_cls (bool, optional): \u5982\u679c\u4e3aTrue，\u5219\u4f7f\u7528\u5355\u7c7b\u8bad\u7ec3，\u9ed8\u8ba4\u4e3aFalse\u3002
            \n classes (list): \u5305\u542b\u7684\u7c7b\u7684\u5217\u8868，\u9ed8\u8ba4\u4e3aNone\u3002
            \n fraction (float): \u4f7f\u7528\u7684\u6570\u636e\u96c6\u7684\u6bd4\u4f8b，\u9ed8\u8ba4\u4e3a1.0（\u4f7f\u7528\u6240\u6709\u6570\u636e）\u3002<\/p>\n

            \u5c5e\u6027:
            \n im_files (list): \u56fe\u50cf\u6587\u4ef6\u8def\u5f84\u5217\u8868\u3002
            \n labels (list): \u6807\u7b7e\u6570\u636e\u5b57\u5178\u5217\u8868\u3002
            \n ni (int): \u6570\u636e\u96c6\u4e2d\u56fe\u50cf\u7684\u6570\u91cf\u3002
            \n ims (list): \u52a0\u8f7d\u7684\u56fe\u50cf\u5217\u8868\u3002
            \n npy_files (list): numpy\u6587\u4ef6\u8def\u5f84\u5217\u8868\u3002
            \n transforms (callable): \u56fe\u50cf\u8f6c\u6362\u51fd\u6570\u3002
            \n """<\/p>\n

            def __init__(self, img_path, imgsz=640, cache=False, augment=True, hyp=DEFAULT_CFG, prefix='', rect=False, batch_size=16, stride=32, pad=0.5, single_cls=False, classes=None, fraction=1.0):
            \n """\u4f7f\u7528\u7ed9\u5b9a\u7684\u914d\u7f6e\u548c\u9009\u9879\u521d\u59cb\u5316BaseDataset\u3002"""
            \n super().__init__()
            \n self.img_path = img_path # \u56fe\u50cf\u8def\u5f84
            \n self.imgsz = imgsz # \u56fe\u50cf\u5927\u5c0f
            \n self.augment = augment # \u662f\u5426\u5e94\u7528\u6570\u636e\u589e\u5f3a
            \n self.single_cls = single_cls # \u662f\u5426\u4f7f\u7528\u5355\u7c7b\u8bad\u7ec3
            \n self.prefix = prefix # \u65e5\u5fd7\u524d\u7f00
            \n self.fraction = fraction # \u4f7f\u7528\u7684\u6570\u636e\u96c6\u6bd4\u4f8b
            \n self.im_files = self.get_img_files(self.img_path) # \u83b7\u53d6\u56fe\u50cf\u6587\u4ef6\u5217\u8868
            \n self.labels = self.get_labels() # \u83b7\u53d6\u6807\u7b7e
            \n self.update_labels(include_class=classes) # \u66f4\u65b0\u6807\u7b7e\u4ee5\u5305\u542b\u6307\u5b9a\u7c7b
            \n self.ni = len(self.labels) # \u6570\u636e\u96c6\u4e2d\u56fe\u50cf\u7684\u6570\u91cf
            \n self.rect = rect # \u662f\u5426\u4f7f\u7528\u77e9\u5f62\u8bad\u7ec3
            \n self.batch_size = batch_size # \u6279\u6b21\u5927\u5c0f
            \n self.stride = stride # \u6b65\u5e45
            \n self.pad = pad # \u586b\u5145
            \n if self.rect:
            \n assert self.batch_size is not None # \u786e\u4fdd\u6279\u6b21\u5927\u5c0f\u5df2\u5b9a\u4e49
            \n self.set_rectangle() # \u8bbe\u7f6e\u77e9\u5f62\u5f62\u72b6<\/p>\n

            # \u7f13\u5b58\u56fe\u50cf
            \n self.buffer = [] # \u7f13\u5b58\u5927\u5c0f = \u6279\u6b21\u5927\u5c0f
            \n self.max_buffer_length = min((self.ni, self.batch_size * 8, 1000)) if self.augment else 0 # \u6700\u5927\u7f13\u5b58\u957f\u5ea6<\/p>\n

            # \u68c0\u67e5\u662f\u5426\u7f13\u5b58\u56fe\u50cf
            \n if cache == 'ram' and not self.check_cache_ram():
            \n cache = False
            \n self.ims, self.im_hw0, self.im_hw = [None] * self.ni, [None] * self.ni, [None] * self.ni # \u521d\u59cb\u5316\u56fe\u50cf\u548c\u5c3a\u5bf8
            \n self.npy_files = [Path(f).with_suffix('.npy') for f in self.im_files] # numpy\u6587\u4ef6\u8def\u5f84
            \n if cache:
            \n self.cache_images(cache) # \u7f13\u5b58\u56fe\u50cf<\/p>\n

            # \u56fe\u50cf\u8f6c\u6362
            \n self.transforms = self.build_transforms(hyp=hyp) # \u6784\u5efa\u56fe\u50cf\u8f6c\u6362<\/p>\n

            def get_img_files(self, img_path):
            \n """\u8bfb\u53d6\u56fe\u50cf\u6587\u4ef6\u3002"""
            \n try:
            \n f = [] # \u56fe\u50cf\u6587\u4ef6\u5217\u8868
            \n for p in img_path if isinstance(img_path, list) else [img_path]:
            \n p = Path(p) # \u5904\u7406\u8def\u5f84
            \n if p.is_dir(): # \u5982\u679c\u662f\u76ee\u5f55
            \n f += glob.glob(str(p \/ '**' \/ '*.*'), recursive=True) # \u83b7\u53d6\u6240\u6709\u56fe\u50cf\u6587\u4ef6
            \n elif p.is_file(): # \u5982\u679c\u662f\u6587\u4ef6
            \n with open(p) as t:
            \n t = t.read().strip().splitlines() # \u8bfb\u53d6\u6587\u4ef6\u5185\u5bb9
            \n parent = str(p.parent) + os.sep
            \n f += [x.replace('.\/', parent) if x.startswith('.\/') else x for x in t] # \u66ff\u6362\u8def\u5f84
            \n else:
            \n raise FileNotFoundError(f'{self.prefix}{p} does not exist') # \u6587\u4ef6\u4e0d\u5b58\u5728
            \n im_files = sorted(x.replace('\/', os.sep) for x in f if x.split('.')[-1].lower() in IMG_FORMATS) # \u8fc7\u6ee4\u56fe\u50cf\u683c\u5f0f
            \n assert im_files, f'{self.prefix}No images found in {img_path}' # \u786e\u4fdd\u627e\u5230\u56fe\u50cf
            \n except Exception as e:
            \n raise FileNotFoundError(f'{self.prefix}Error loading data from {img_path}\\\\n{HELP_URL}') from e
            \n if self.fraction < 1:
            \n im_files = im_files[:round(len(im_files) * self.fraction)] # \u6839\u636e\u6bd4\u4f8b\u9009\u62e9\u56fe\u50cf
            \n return im_files<\/p>\n

            def load_image(self, i, rect_mode=True):
            \n """\u4ece\u6570\u636e\u96c6\u4e2d\u52a0\u8f7d\u4e00\u5f20\u56fe\u50cf，\u8fd4\u56de\u56fe\u50cf\u53ca\u5176\u539f\u59cb\u548c\u8c03\u6574\u540e\u7684\u5c3a\u5bf8\u3002"""
            \n im, f, fn = self.ims[i], self.im_files[i], self.npy_files[i] # \u83b7\u53d6\u56fe\u50cf\u3001\u6587\u4ef6\u8def\u5f84\u548cnumpy\u6587\u4ef6\u8def\u5f84
            \n if im is None: # \u5982\u679c\u672a\u7f13\u5b58
            \n if fn.exists(): # \u5982\u679c\u5b58\u5728numpy\u6587\u4ef6
            \n try:
            \n im = np.load(fn) # \u52a0\u8f7dnumpy\u6587\u4ef6
            \n except Exception as e:
            \n LOGGER.warning(f'{self.prefix}WARNING \u26a0\ufe0f Removing corrupt *.npy image file {fn} due to: {e}') # \u8b66\u544a\u5e76\u5220\u9664\u635f\u574f\u7684\u6587\u4ef6
            \n Path(fn).unlink(missing_ok=True) # \u5220\u9664\u635f\u574f\u7684\u6587\u4ef6
            \n im = cv2.imread(f) # \u8bfb\u53d6\u56fe\u50cf
            \n else: # \u8bfb\u53d6\u56fe\u50cf
            \n im = cv2.imread(f) # \u8bfb\u53d6\u56fe\u50cf
            \n if im is None:
            \n raise FileNotFoundError(f'Image Not Found {f}') # \u56fe\u50cf\u672a\u627e\u5230<\/p>\n

            h0, w0 = im.shape[:2] # \u83b7\u53d6\u539f\u59cb\u9ad8\u5ea6\u548c\u5bbd\u5ea6
            \n if rect_mode: # \u5982\u679c\u4f7f\u7528\u77e9\u5f62\u6a21\u5f0f
            \n r = self.imgsz \/ max(h0, w0) # \u8ba1\u7b97\u7f29\u653e\u6bd4\u4f8b
            \n if r != 1: # \u5982\u679c\u5c3a\u5bf8\u4e0d\u76f8\u7b49
            \n w, h = (min(math.ceil(w0 * r), self.imgsz), min(math.ceil(h0 * r), self.imgsz)) # \u8c03\u6574\u5c3a\u5bf8
            \n im = cv2.resize(im, (w, h), interpolation=cv2.INTER_LINEAR) # \u8c03\u6574\u56fe\u50cf\u5927\u5c0f
            \n elif not (h0 == w0 == self.imgsz): # \u5982\u679c\u4e0d\u662f\u6b63\u65b9\u5f62
            \n im = cv2.resize(im, (self.imgsz, self.imgsz), interpolation=cv2.INTER_LINEAR) # \u62c9\u4f38\u56fe\u50cf\u5230\u6b63\u65b9\u5f62<\/p>\n

            # \u5982\u679c\u8fdb\u884c\u6570\u636e\u589e\u5f3a，\u5219\u6dfb\u52a0\u5230\u7f13\u5b58
            \n if self.augment:
            \n self.ims[i], self.im_hw0[i], self.im_hw[i] = im, (h0, w0), im.shape[:2] # \u5b58\u50a8\u56fe\u50cf\u548c\u5c3a\u5bf8
            \n self.buffer.append(i) # \u6dfb\u52a0\u5230\u7f13\u5b58
            \n if len(self.buffer) >= self.max_buffer_length: # \u5982\u679c\u7f13\u5b58\u5df2\u6ee1
            \n j = self.buffer.pop(0) # \u79fb\u9664\u6700\u65e7\u7684\u56fe\u50cf
            \n self.ims[j], self.im_hw0[j], self.im_hw[j] = None, None, None # \u6e05\u7a7a\u7f13\u5b58<\/p>\n

            return im, (h0, w0), im.shape[:2] # \u8fd4\u56de\u56fe\u50cf\u53ca\u5176\u5c3a\u5bf8<\/p>\n

            return self.ims[i], self.im_hw0[i], self.im_hw[i] # \u8fd4\u56de\u7f13\u5b58\u7684\u56fe\u50cf\u548c\u5c3a\u5bf8<\/p>\n

            def __getitem__(self, index):
            \n """\u8fd4\u56de\u7ed9\u5b9a\u7d22\u5f15\u7684\u8f6c\u6362\u6807\u7b7e\u4fe1\u606f\u3002"""
            \n return self.transforms(self.get_image_and_label(index)) # \u5e94\u7528\u8f6c\u6362\u5e76\u8fd4\u56de\u6807\u7b7e\u4fe1\u606f<\/p>\n

            def __len__(self):
            \n """\u8fd4\u56de\u6570\u636e\u96c6\u4e2d\u6807\u7b7e\u5217\u8868\u7684\u957f\u5ea6\u3002"""
            \n return len(self.labels) # \u8fd4\u56de\u6807\u7b7e\u6570\u91cf<\/p>\n

            def build_transforms(self, hyp=None):
            \n """
            \n \u7528\u6237\u53ef\u4ee5\u5728\u6b64\u81ea\u5b9a\u4e49\u589e\u5f3a\u3002<\/p>\n

            \u793a\u4f8b:
            \n ```python
            \n if self.augment:
            \n # \u8bad\u7ec3\u8f6c\u6362
            \n return Compose([])
            \n else:
            \n # \u9a8c\u8bc1\u8f6c\u6362
            \n return Compose([])
            \n ```
            \n """
            \n raise NotImplementedError # \u9700\u8981\u7528\u6237\u5b9e\u73b0<\/p>\n

            def get_labels(self):
            \n """
            \n \u7528\u6237\u53ef\u4ee5\u5728\u6b64\u81ea\u5b9a\u4e49\u6807\u7b7e\u683c\u5f0f\u3002<\/p>\n

            \u6ce8\u610f:
            \n \u786e\u4fdd\u8f93\u51fa\u662f\u4e00\u4e2a\u5b57\u5178，\u5305\u542b\u4ee5\u4e0b\u952e:
            \n ```python
            \n dict(
            \n im_file=im_file,
            \n shape=shape, # \u683c\u5f0f: (\u9ad8\u5ea6, \u5bbd\u5ea6)
            \n cls=cls,
            \n bboxes=bboxes, # xywh
            \n segments=segments, # xy
            \n keypoints=keypoints, # xy
            \n normalized=True, # \u6216 False
            \n bbox_format="xyxy", # \u6216 xywh, ltwh
            \n )
            \n ```
            \n """
            \n raise NotImplementedError # \u9700\u8981\u7528\u6237\u5b9e\u73b0<\/p>\n

            \u4ee3\u7801\u6838\u5fc3\u90e8\u5206\u89e3\u91ca<\/h4>\n
          • \n

            BaseDataset\u7c7b: \u8fd9\u662f\u4e00\u4e2a\u7ee7\u627f\u81eaDataset\u7684\u57fa\u7840\u7c7b，\u7528\u4e8e\u52a0\u8f7d\u548c\u5904\u7406\u56fe\u50cf\u6570\u636e\u3002\u5b83\u63d0\u4f9b\u4e86\u56fe\u50cf\u6587\u4ef6\u7684\u8bfb\u53d6\u3001\u6807\u7b7e\u7684\u7ba1\u7406\u3001\u56fe\u50cf\u7684\u52a0\u8f7d\u548c\u8f6c\u6362\u7b49\u529f\u80fd\u3002<\/p>\n<\/li>\n

          • \n

            \u521d\u59cb\u5316\u65b9\u6cd5: \u5728\u521d\u59cb\u5316\u65f6，\u7c7b\u4f1a\u63a5\u6536\u591a\u79cd\u53c2\u6570，\u5305\u62ec\u56fe\u50cf\u8def\u5f84\u3001\u56fe\u50cf\u5927\u5c0f\u3001\u662f\u5426\u7f13\u5b58\u3001\u6570\u636e\u589e\u5f3a\u7b49\u3002\u5b83\u4f1a\u8c03\u7528\u76f8\u5173\u65b9\u6cd5\u6765\u83b7\u53d6\u56fe\u50cf\u6587\u4ef6\u548c\u6807\u7b7e，\u5e76\u8fdb\u884c\u5fc5\u8981\u7684\u8bbe\u7f6e\u3002<\/p>\n<\/li>\n

          • \n

            \u56fe\u50cf\u6587\u4ef6\u8bfb\u53d6: get_img_files\u65b9\u6cd5\u7528\u4e8e\u8bfb\u53d6\u6307\u5b9a\u8def\u5f84\u4e0b\u7684\u56fe\u50cf\u6587\u4ef6，\u5e76\u8fd4\u56de\u4e00\u4e2a\u56fe\u50cf\u6587\u4ef6\u8def\u5f84\u7684\u5217\u8868\u3002<\/p>\n<\/li>\n

          • \n

            \u56fe\u50cf\u52a0\u8f7d: load_image\u65b9\u6cd5\u7528\u4e8e\u52a0\u8f7d\u5355\u5f20\u56fe\u50cf，\u652f\u6301\u4ece\u7f13\u5b58\u4e2d\u8bfb\u53d6\u6216\u4ece\u6587\u4ef6\u4e2d\u8bfb\u53d6，\u5e76\u6839\u636e\u9700\u8981\u8c03\u6574\u56fe\u50cf\u5927\u5c0f\u3002<\/p>\n<\/li>\n

          • \n

            \u6570\u636e\u589e\u5f3a\u548c\u8f6c\u6362: __getitem__\u65b9\u6cd5\u8fd4\u56de\u7ed9\u5b9a\u7d22\u5f15\u7684\u56fe\u50cf\u548c\u6807\u7b7e\u4fe1\u606f，\u5e76\u5e94\u7528\u6570\u636e\u589e\u5f3a\u3002build_transforms\u65b9\u6cd5\u7528\u4e8e\u5b9a\u4e49\u56fe\u50cf\u7684\u8f6c\u6362\u64cd\u4f5c\u3002<\/p>\n<\/li>\n

          • \n

            \u6807\u7b7e\u7ba1\u7406: get_labels\u65b9\u6cd5\u7528\u4e8e\u83b7\u53d6\u6807\u7b7e\u4fe1\u606f，\u7528\u6237\u53ef\u4ee5\u81ea\u5b9a\u4e49\u6807\u7b7e\u7684\u683c\u5f0f\u3002<\/p>\n<\/li>\n

            \u8fd9\u4e2a\u7c7b\u662f\u6570\u636e\u96c6\u5904\u7406\u7684\u57fa\u7840，\u540e\u7eed\u53ef\u4ee5\u5728\u6b64\u57fa\u7840\u4e0a\u6269\u5c55\u66f4\u591a\u529f\u80fd，\u4f8b\u5982\u81ea\u5b9a\u4e49\u6570\u636e\u589e\u5f3a\u7b56\u7565\u6216\u6807\u7b7e\u683c\u5f0f\u3002```
            \n\u8fd9\u4e2a\u7a0b\u5e8f\u6587\u4ef6\u5b9a\u4e49\u4e86\u4e00\u4e2a\u540d\u4e3a BaseDataset \u7684\u7c7b，\u7ee7\u627f\u81ea PyTorch \u7684 Dataset \u7c7b，\u4e3b\u8981\u7528\u4e8e\u52a0\u8f7d\u548c\u5904\u7406\u56fe\u50cf\u6570\u636e，\u7279\u522b\u662f\u7528\u4e8e\u8bad\u7ec3 YOLO（You Only Look Once）\u6a21\u578b\u7684\u56fe\u50cf\u6570\u636e\u96c6\u3002\u8be5\u7c7b\u63d0\u4f9b\u4e86\u4e00\u7cfb\u5217\u65b9\u6cd5\u548c\u5c5e\u6027，\u4ee5\u4fbf\u4e8e\u7ba1\u7406\u56fe\u50cf\u6587\u4ef6\u3001\u6807\u7b7e\u3001\u6570\u636e\u589e\u5f3a\u7b49\u529f\u80fd\u3002<\/p>\n

            \u5728\u521d\u59cb\u5316\u65b9\u6cd5 __init__ \u4e2d，\u7528\u6237\u53ef\u4ee5\u901a\u8fc7\u4f20\u5165\u4e0d\u540c\u7684\u53c2\u6570\u6765\u914d\u7f6e\u6570\u636e\u96c6\u7684\u52a0\u8f7d\u65b9\u5f0f，\u5305\u62ec\u56fe\u50cf\u8def\u5f84\u3001\u56fe\u50cf\u5927\u5c0f\u3001\u662f\u5426\u7f13\u5b58\u56fe\u50cf\u3001\u662f\u5426\u5e94\u7528\u6570\u636e\u589e\u5f3a\u3001\u8d85\u53c2\u6570\u3001\u6279\u91cf\u5927\u5c0f\u7b49\u3002\u521d\u59cb\u5316\u8fc7\u7a0b\u4e2d，\u7c7b\u4f1a\u8c03\u7528 get_img_files \u65b9\u6cd5\u6765\u8bfb\u53d6\u56fe\u50cf\u6587\u4ef6，\u5e76\u901a\u8fc7 get_labels \u65b9\u6cd5\u83b7\u53d6\u76f8\u5e94\u7684\u6807\u7b7e\u6570\u636e\u3002\u7c7b\u8fd8\u4f1a\u6839\u636e\u7528\u6237\u7684\u8bbe\u7f6e\u6765\u51b3\u5b9a\u662f\u5426\u4f7f\u7528\u77e9\u5f62\u8bad\u7ec3\u6a21\u5f0f\u3001\u7f13\u5b58\u56fe\u50cf\u7b49\u3002<\/p>\n

            get_img_files \u65b9\u6cd5\u8d1f\u8d23\u4ece\u6307\u5b9a\u8def\u5f84\u8bfb\u53d6\u56fe\u50cf\u6587\u4ef6，\u53ef\u4ee5\u5904\u7406\u6587\u4ef6\u5939\u548c\u6587\u4ef6\u5217\u8868，\u652f\u6301\u9012\u5f52\u67e5\u627e，\u5e76\u786e\u4fdd\u53ea\u8fd4\u56de\u7b26\u5408\u6307\u5b9a\u683c\u5f0f\u7684\u56fe\u50cf\u6587\u4ef6\u3002\u82e5\u6307\u5b9a\u4e86\u6570\u636e\u96c6\u7684\u4f7f\u7528\u6bd4\u4f8b（fraction），\u5219\u4f1a\u6839\u636e\u8be5\u6bd4\u4f8b\u8fd4\u56de\u90e8\u5206\u56fe\u50cf\u6587\u4ef6\u3002<\/p>\n

            update_labels \u65b9\u6cd5\u7528\u4e8e\u66f4\u65b0\u6807\u7b7e\u4fe1\u606f，\u4ee5\u4fbf\u53ea\u4fdd\u7559\u7528\u6237\u6307\u5b9a\u7684\u7c7b\u522b\u3002\u8be5\u65b9\u6cd5\u4f1a\u6839\u636e\u4f20\u5165\u7684\u7c7b\u522b\u5217\u8868\u8fc7\u6ee4\u6807\u7b7e，\u5e76\u5904\u7406\u5355\u7c7b\u8bad\u7ec3\u7684\u60c5\u51b5\u3002<\/p>\n

            load_image \u65b9\u6cd5\u7528\u4e8e\u52a0\u8f7d\u6307\u5b9a\u7d22\u5f15\u7684\u56fe\u50cf\u3002\u5982\u679c\u56fe\u50cf\u672a\u7f13\u5b58，\u5219\u4f1a\u5c1d\u8bd5\u4ece\u78c1\u76d8\u8bfb\u53d6\u56fe\u50cf\u6216\u5bf9\u5e94\u7684 .npy \u6587\u4ef6\u3002\u52a0\u8f7d\u7684\u56fe\u50cf\u4f1a\u6839\u636e\u7528\u6237\u7684\u8bbe\u7f6e\u8fdb\u884c\u8c03\u6574（\u5982\u957f\u8fb9\u7f29\u653e\u5230\u6307\u5b9a\u5927\u5c0f），\u5e76\u4e14\u5728\u6570\u636e\u589e\u5f3a\u7684\u60c5\u51b5\u4e0b，\u4f1a\u5c06\u56fe\u50cf\u6dfb\u52a0\u5230\u7f13\u51b2\u533a\u3002<\/p>\n

            cache_images \u548c cache_images_to_disk \u65b9\u6cd5\u7528\u4e8e\u5c06\u56fe\u50cf\u7f13\u5b58\u5230\u5185\u5b58\u6216\u78c1\u76d8，\u4ee5\u63d0\u9ad8\u540e\u7eed\u52a0\u8f7d\u7684\u901f\u5ea6\u3002check_cache_ram \u65b9\u6cd5\u5219\u7528\u4e8e\u68c0\u67e5\u5f53\u524d\u7cfb\u7edf\u7684\u5185\u5b58\u662f\u5426\u8db3\u591f\u7f13\u5b58\u6574\u4e2a\u6570\u636e\u96c6\u3002<\/p>\n

            set_rectangle \u65b9\u6cd5\u7528\u4e8e\u8bbe\u7f6e YOLO \u68c0\u6d4b\u7684\u8fb9\u754c\u6846\u5f62\u72b6\u4e3a\u77e9\u5f62，\u4ee5\u4fbf\u4e8e\u5728\u77e9\u5f62\u8bad\u7ec3\u6a21\u5f0f\u4e0b\u8fdb\u884c\u8bad\u7ec3\u3002<\/p>\n

            __getitem__ \u65b9\u6cd5\u662f\u6570\u636e\u96c6\u7684\u6838\u5fc3\u65b9\u6cd5\u4e4b\u4e00，\u7528\u4e8e\u6839\u636e\u7d22\u5f15\u8fd4\u56de\u7ecf\u8fc7\u53d8\u6362\u7684\u56fe\u50cf\u548c\u6807\u7b7e\u4fe1\u606f\u3002get_image_and_label \u65b9\u6cd5\u5219\u8d1f\u8d23\u83b7\u53d6\u6307\u5b9a\u7d22\u5f15\u7684\u56fe\u50cf\u548c\u6807\u7b7e，\u5e76\u8fdb\u884c\u5fc5\u8981\u7684\u5904\u7406\u3002<\/p>\n

            __len__ \u65b9\u6cd5\u8fd4\u56de\u6570\u636e\u96c6\u4e2d\u6807\u7b7e\u7684\u6570\u91cf\u3002<\/p>\n

            \u6b64\u5916，build_transforms \u548c get_labels \u65b9\u6cd5\u662f\u7559\u7ed9\u7528\u6237\u81ea\u5b9a\u4e49\u7684\u63a5\u53e3，\u5141\u8bb8\u7528\u6237\u6839\u636e\u9700\u8981\u5b9e\u73b0\u81ea\u5df1\u7684\u6570\u636e\u589e\u5f3a\u548c\u6807\u7b7e\u683c\u5f0f\u3002<\/p>\n

            \u603b\u4f53\u800c\u8a00，\u8fd9\u4e2a\u7c7b\u4e3a\u56fe\u50cf\u6570\u636e\u96c6\u7684\u52a0\u8f7d\u3001\u5904\u7406\u548c\u589e\u5f3a\u63d0\u4f9b\u4e86\u7075\u6d3b\u7684\u6846\u67b6，\u9002\u7528\u4e8e\u8bad\u7ec3\u8ba1\u7b97\u673a\u89c6\u89c9\u6a21\u578b，\u5c24\u5176\u662f YOLO \u7cfb\u5217\u6a21\u578b\u3002<\/p>\n

            ```python
            \nimport torch # \u5bfc\u5165PyTorch\u5e93<\/p>\n

            # \u5b9a\u4e49YOLO\u6a21\u578b\u7c7b
            \nclass YOLO:
            \n def __init__(self, model_path):
            \n # \u52a0\u8f7d\u9884\u8bad\u7ec3\u7684YOLO\u6a21\u578b
            \n self.model = torch.hub.load('ultralytics\/yolov5', 'yolov5s', pretrained=True)<\/p>\n

            def detect(self, img):
            \n # \u5bf9\u8f93\u5165\u56fe\u50cf\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b
            \n results = self.model(img) # \u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u63a8\u7406
            \n return results # \u8fd4\u56de\u68c0\u6d4b\u7ed3\u679c<\/p>\n

            # \u4e3b\u7a0b\u5e8f
            \nif __name__ == "__main__":
            \n # \u521b\u5efaYOLO\u5bf9\u8c61，\u52a0\u8f7d\u6a21\u578b
            \n yolo = YOLO('yolov5s.pt')<\/p>\n

            # \u52a0\u8f7d\u5f85\u68c0\u6d4b\u7684\u56fe\u50cf
            \n img = 'path\/to\/image.jpg' # \u66ff\u6362\u4e3a\u5b9e\u9645\u56fe\u50cf\u8def\u5f84<\/p>\n

            # \u8fdb\u884c\u76ee\u6807\u68c0\u6d4b
            \n results = yolo.detect(img)<\/p>\n

            # \u6253\u5370\u68c0\u6d4b\u7ed3\u679c
            \n print(results) # \u8f93\u51fa\u68c0\u6d4b\u5230\u7684\u76ee\u6807\u4fe1\u606f<\/p>\n

            \u4ee3\u7801\u8be6\u7ec6\u6ce8\u91ca：<\/h4>\n
          • \n

            \u5bfc\u5165\u5e93：<\/p>\n

            import<\/span> torch # \u5bfc\u5165PyTorch\u5e93，\u7528\u4e8e\u6df1\u5ea6\u5b66\u4e60\u6a21\u578b\u7684\u52a0\u8f7d\u548c\u63a8\u7406<\/span><\/p>\n<\/li>\n

          • \n

            \u5b9a\u4e49YOLO\u6a21\u578b\u7c7b：<\/p>\n

            class<\/span> YOLO<\/span>:<\/span><\/p>\n

              \n
            • \u8fd9\u662f\u4e00\u4e2aYOLO\u6a21\u578b\u7684\u7c7b，\u8d1f\u8d23\u52a0\u8f7d\u6a21\u578b\u548c\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u3002<\/li>\n<\/ul>\n<\/li>\n
            • \n

              \u521d\u59cb\u5316\u65b9\u6cd5：<\/p>\n

              def<\/span> __init__<\/span>(<\/span>self,<\/span> model_path)<\/span>:<\/span><\/p>\n

                \n
              • \u6784\u9020\u51fd\u6570，\u7528\u4e8e\u521d\u59cb\u5316YOLO\u5bf9\u8c61\u5e76\u52a0\u8f7d\u6a21\u578b\u3002<\/li>\n<\/ul>\n<\/li>\n
              • \n

                \u52a0\u8f7d\u6a21\u578b：<\/p>\n

                self.<\/span>model =<\/span> torch.<\/span>hub.<\/span>load(<\/span>'ultralytics\/yolov5'<\/span>,<\/span> 'yolov5s'<\/span>,<\/span> pretrained=<\/span>True<\/span>)<\/span><\/p>\n

                  \n
                • \u4f7f\u7528torch.hub\u4eceUltralytics\u7684YOLOv5\u5e93\u4e2d\u52a0\u8f7d\u9884\u8bad\u7ec3\u7684YOLOv5s\u6a21\u578b\u3002<\/li>\n<\/ul>\n<\/li>\n
                • \n

                  \u76ee\u6807\u68c0\u6d4b\u65b9\u6cd5：<\/p>\n

                  def<\/span> detect<\/span>(<\/span>self,<\/span> img)<\/span>:<\/span><\/p>\n

                    \n
                  • \u5b9a\u4e49\u4e00\u4e2a\u65b9\u6cd5，\u7528\u4e8e\u5bf9\u8f93\u5165\u56fe\u50cf\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u3002<\/li>\n<\/ul>\n<\/li>\n
                  • \n

                    \u6a21\u578b\u63a8\u7406：<\/p>\n

                    results =<\/span> self.<\/span>model(<\/span>img)<\/span> # \u4f7f\u7528\u6a21\u578b\u8fdb\u884c\u63a8\u7406<\/span><\/p>\n

                      \n
                    • \u5c06\u8f93\u5165\u56fe\u50cf\u4f20\u9012\u7ed9\u6a21\u578b，\u8fdb\u884c\u63a8\u7406，\u8fd4\u56de\u68c0\u6d4b\u7ed3\u679c\u3002<\/li>\n<\/ul>\n<\/li>\n
                    • \n

                      \u4e3b\u7a0b\u5e8f：<\/p>\n

                      if<\/span> __name__ ==<\/span> "__main__"<\/span>:<\/span><\/p>\n

                        \n
                      • \u786e\u4fdd\u53ea\u6709\u5728\u76f4\u63a5\u8fd0\u884c\u8be5\u811a\u672c\u65f6\u624d\u4f1a\u6267\u884c\u4ee5\u4e0b\u4ee3\u7801\u3002<\/li>\n<\/ul>\n<\/li>\n
                      • \n

                        \u521b\u5efaYOLO\u5bf9\u8c61：<\/p>\n

                        yolo =<\/span> YOLO(<\/span>'yolov5s.pt'<\/span>)<\/span><\/p>\n

                          \n
                        • \u5b9e\u4f8b\u5316YOLO\u7c7b，\u52a0\u8f7d\u6a21\u578b\u3002<\/li>\n<\/ul>\n<\/li>\n
                        • \n

                          \u52a0\u8f7d\u5f85\u68c0\u6d4b\u56fe\u50cf：<\/p>\n

                          img =<\/span> 'path\/to\/image.jpg'<\/span> # \u66ff\u6362\u4e3a\u5b9e\u9645\u56fe\u50cf\u8def\u5f84<\/span><\/p>\n

                            \n
                          • \u6307\u5b9a\u5f85\u68c0\u6d4b\u7684\u56fe\u50cf\u8def\u5f84\u3002<\/li>\n<\/ul>\n<\/li>\n
                          • \n

                            \u8fdb\u884c\u76ee\u6807\u68c0\u6d4b：<\/p>\n

                            results =<\/span> yolo.<\/span>detect(<\/span>img)<\/span><\/p>\n

                              \n
                            • \u8c03\u7528\u68c0\u6d4b\u65b9\u6cd5，\u83b7\u53d6\u68c0\u6d4b\u7ed3\u679c\u3002<\/li>\n<\/ul>\n<\/li>\n
                            • \n

                              \u6253\u5370\u68c0\u6d4b\u7ed3\u679c：<\/p>\n

                              print<\/span>(<\/span>results)<\/span> # \u8f93\u51fa\u68c0\u6d4b\u5230\u7684\u76ee\u6807\u4fe1\u606f<\/span><\/p>\n

                                \n
                              • \u8f93\u51fa\u68c0\u6d4b\u5230\u7684\u76ee\u6807\u4fe1\u606f，\u5305\u62ec\u7c7b\u522b\u3001\u7f6e\u4fe1\u5ea6\u548c\u8fb9\u754c\u6846\u7b49\u3002<\/li>\n<\/ul>\n<\/li>\n

                                \u8fd9\u4e2a\u793a\u4f8b\u4ee3\u7801\u5c55\u793a\u4e86YOLO\u6a21\u578b\u7684\u57fa\u672c\u7ed3\u6784\u548c\u4f7f\u7528\u65b9\u6cd5，\u6ce8\u91ca\u8be6\u7ec6\u89e3\u91ca\u4e86\u6bcf\u4e00\u90e8\u5206\u7684\u529f\u80fd\u3002```
                                \n\u8fd9\u4e2a\u6587\u4ef6\u662fUltralytics YOLO\u9879\u76ee\u7684\u4e00\u90e8\u5206，\u5c5e\u4e8e\u4e00\u4e2a\u5f00\u6e90\u7684\u8ba1\u7b97\u673a\u89c6\u89c9\u6a21\u578b，\u4e3b\u8981\u7528\u4e8e\u76ee\u6807\u68c0\u6d4b\u4efb\u52a1\u3002\u6587\u4ef6\u7684\u5f00\u5934\u5305\u542b\u4e86\u4e00\u4e2a\u6ce8\u91ca，\u6807\u660e\u4e86\u9879\u76ee\u7684\u540d\u79f0\u201cUltralytics YOLO\u201d\u4ee5\u53ca\u5176\u4f7f\u7528\u7684\u8bb8\u53ef\u8bc1\u7c7b\u578b\u201cAGPL-3.0\u201d\u3002AGPL-3.0\u662f\u4e00\u79cd\u5f00\u6e90\u8bb8\u53ef\u8bc1，\u8981\u6c42\u4efb\u4f55\u5bf9\u8be5\u8f6f\u4ef6\u7684\u4fee\u6539\u548c\u5206\u53d1\u90fd\u5fc5\u987b\u4ee5\u76f8\u540c\u7684\u8bb8\u53ef\u8bc1\u8fdb\u884c\u3002<\/p>\n

                                \u867d\u7136\u6587\u4ef6\u7684\u5177\u4f53\u4ee3\u7801\u6ca1\u6709\u63d0\u4f9b，\u4f46\u901a\u5e38\u5728__init__.py\u6587\u4ef6\u4e2d，\u5f00\u53d1\u8005\u4f1a\u521d\u59cb\u5316\u4e00\u4e2a\u5305，\u4f7f\u5f97\u8be5\u76ee\u5f55\u53ef\u4ee5\u88ab\u89c6\u4e3a\u4e00\u4e2aPython\u5305\u3002\u8fd9\u4e2a\u6587\u4ef6\u53ef\u80fd\u4f1a\u5305\u542b\u4e00\u4e9b\u5bfc\u5165\u8bed\u53e5，\u5bfc\u5165\u8be5\u5305\u4e2d\u7684\u5176\u4ed6\u6a21\u5757\u6216\u7c7b，\u6216\u8005\u5b9a\u4e49\u4e00\u4e9b\u516c\u5171\u63a5\u53e3，\u4ee5\u4fbf\u7528\u6237\u53ef\u4ee5\u65b9\u4fbf\u5730\u4f7f\u7528\u8be5\u5305\u7684\u529f\u80fd\u3002<\/p>\n

                                \u5728YOLO\u6a21\u578b\u7684\u4e0a\u4e0b\u6587\u4e2d，\u8fd9\u4e2a\u6587\u4ef6\u53ef\u80fd\u6d89\u53ca\u5230\u4e00\u4e9b\u4e0e\u6a21\u578b\u76f8\u5173\u7684\u5de5\u5177\u51fd\u6570\u3001\u7c7b\u6216\u914d\u7f6e\u8bbe\u7f6e，\u5e2e\u52a9\u7528\u6237\u66f4\u597d\u5730\u4f7f\u7528YOLO\u8fdb\u884c\u76ee\u6807\u68c0\u6d4b\u3002\u603b\u7684\u6765\u8bf4，\u8fd9\u4e2a\u6587\u4ef6\u662fUltralytics YOLO\u9879\u76ee\u7ed3\u6784\u7684\u4e00\u90e8\u5206，\u8d77\u5230\u7ec4\u7ec7\u548c\u7ba1\u7406\u4ee3\u7801\u7684\u4f5c\u7528\u3002<\/p>\n

                                \u6e90\u7801\u6587\u4ef6<\/h2>\n

                                \"\u5728\u8fd9\u91cc\u63d2\u5165\u56fe\u7247\u63cf\u8ff0\"<\/p>\n

                                \u6e90\u7801\u83b7\u53d6<\/h2>\n

                                \u6b22\u8fce\u5927\u5bb6\u70b9\u8d5e\u3001\u6536\u85cf\u3001\u5173\u6ce8\u3001\u8bc4\u8bba\u5566 \u3001\u67e5\u770b👇🏻\u83b7\u53d6\u8054\u7cfb\u65b9\u5f0f👇🏻<\/p>\n","protected":false},"excerpt":{"rendered":"

                                \u80cc\u666f\u610f\u4e49
                                \n\u9f8b\u9f7f\u4f5c\u4e3a\u4e00\u79cd\u5e38\u89c1\u7684\u53e3\u8154\u75be\u75c5，\u5f71\u54cd\u7740\u5168\u7403\u6570\u4ebf\u4eba\u7684\u53e3\u8154\u5065\u5eb7\u3002\u6839\u636e\u4e16\u754c\u536b\u751f\u7ec4\u7ec7\u7684\u7edf\u8ba1，\u9f8b\u9f7f\u7684\u53d1\u75c5\u7387\u5728\u513f\u7ae5\u548c\u9752\u5c11\u5e74\u4e2d\u5c24\u4e3a\u663e\u8457，\u4e14\u968f\u7740\u5e74\u9f84\u7684\u589e\u957f，\u6210\u4eba\u548c\u8001\u5e74\u4eba\u4e5f\u9762\u4e34\u7740\u9f8b\u9f7f\u7684\u5a01\u80c1\u3002\u9f8b\u9f7f\u4e0d\u4ec5\u4f1a\u5bfc\u81f4\u75bc\u75db\u548c\u4e0d\u9002，\u8fd8\u53ef\u80fd\u5f15\u53d1\u66f4\u4e3a\u4e25\u91cd\u7684\u53e3\u8154\u5e76\u53d1\u75c7，\u5982\u7259\u9ad3\u708e\u3001\u6839\u5c16\u5468\u56f4\u708e\u7b49\u3002\u56e0\u6b64，\u65e9\u671f\u7684\u9f8b\u9f7f\u68c0\u6d4b\u4e0e\u8bca\u65ad\u663e\u5f97\u5c24\u4e3a\u91cd\u8981\u3002\u4f20\u7edf\u7684\u9f8b\u9f7f\u8bca\u65ad\u65b9\u6cd5\u4e3b\u8981\u4f9d\u8d56\u4e8e\u533b\u751f\u7684\u89c6\u89c9\u68c0\u67e5\u548cX\u5149\u5f71\u50cf，\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u4ec5\u8017<\/p>\n","protected":false},"author":2,"featured_media":72675,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1],"tags":[7646,156,2395,87,5134,5281,523],"topic":[],"class_list":["post-72684","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-server","tag-7646","tag-yolo","tag-2395","tag-87","tag-5134","tag-5281","tag-523"],"yoast_head":"\n\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3<\/title>\n<meta name=\"robots\" content=\"index, follow, max-snippet:-1, max-image-preview:large, max-video-preview:-1\" \/>\n<link rel=\"canonical\" href=\"https:\/\/www.wsisp.com\/helps\/72684.html\" \/>\n<meta property=\"og:locale\" content=\"zh_CN\" \/>\n<meta property=\"og:type\" content=\"article\" \/>\n<meta property=\"og:title\" content=\"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"og:description\" content=\"\u80cc\u666f\u610f\u4e49 \u9f8b\u9f7f\u4f5c\u4e3a\u4e00\u79cd\u5e38\u89c1\u7684\u53e3\u8154\u75be\u75c5，\u5f71\u54cd\u7740\u5168\u7403\u6570\u4ebf\u4eba\u7684\u53e3\u8154\u5065\u5eb7\u3002\u6839\u636e\u4e16\u754c\u536b\u751f\u7ec4\u7ec7\u7684\u7edf\u8ba1，\u9f8b\u9f7f\u7684\u53d1\u75c5\u7387\u5728\u513f\u7ae5\u548c\u9752\u5c11\u5e74\u4e2d\u5c24\u4e3a\u663e\u8457，\u4e14\u968f\u7740\u5e74\u9f84\u7684\u589e\u957f，\u6210\u4eba\u548c\u8001\u5e74\u4eba\u4e5f\u9762\u4e34\u7740\u9f8b\u9f7f\u7684\u5a01\u80c1\u3002\u9f8b\u9f7f\u4e0d\u4ec5\u4f1a\u5bfc\u81f4\u75bc\u75db\u548c\u4e0d\u9002，\u8fd8\u53ef\u80fd\u5f15\u53d1\u66f4\u4e3a\u4e25\u91cd\u7684\u53e3\u8154\u5e76\u53d1\u75c7，\u5982\u7259\u9ad3\u708e\u3001\u6839\u5c16\u5468\u56f4\u708e\u7b49\u3002\u56e0\u6b64，\u65e9\u671f\u7684\u9f8b\u9f7f\u68c0\u6d4b\u4e0e\u8bca\u65ad\u663e\u5f97\u5c24\u4e3a\u91cd\u8981\u3002\u4f20\u7edf\u7684\u9f8b\u9f7f\u8bca\u65ad\u65b9\u6cd5\u4e3b\u8981\u4f9d\u8d56\u4e8e\u533b\u751f\u7684\u89c6\u89c9\u68c0\u67e5\u548cX\u5149\u5f71\u50cf，\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u4ec5\u8017\" \/>\n<meta property=\"og:url\" content=\"https:\/\/www.wsisp.com\/helps\/72684.html\" \/>\n<meta property=\"og:site_name\" content=\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\" \/>\n<meta property=\"article:published_time\" content=\"2026-02-06T03:50:01+00:00\" \/>\n<meta property=\"og:image\" content=\"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/02\/20260206034950-698564de8d177.jpg\" \/>\n<meta name=\"author\" content=\"admin\" \/>\n<meta name=\"twitter:card\" content=\"summary_large_image\" \/>\n<meta name=\"twitter:label1\" content=\"\u4f5c\u8005\" \/>\n\t<meta name=\"twitter:data1\" content=\"admin\" \/>\n\t<meta name=\"twitter:label2\" content=\"\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4\" \/>\n\t<meta name=\"twitter:data2\" content=\"13 \u5206\" \/>\n<script type=\"application\/ld+json\" class=\"yoast-schema-graph\">{\"@context\":\"https:\/\/schema.org\",\"@graph\":[{\"@type\":\"WebPage\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/72684.html\",\"url\":\"https:\/\/www.wsisp.com\/helps\/72684.html\",\"name\":\"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"isPartOf\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\"},\"datePublished\":\"2026-02-06T03:50:01+00:00\",\"dateModified\":\"2026-02-06T03:50:01+00:00\",\"author\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\"},\"breadcrumb\":{\"@id\":\"https:\/\/www.wsisp.com\/helps\/72684.html#breadcrumb\"},\"inLanguage\":\"zh-Hans\",\"potentialAction\":[{\"@type\":\"ReadAction\",\"target\":[\"https:\/\/www.wsisp.com\/helps\/72684.html\"]}]},{\"@type\":\"BreadcrumbList\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/72684.html#breadcrumb\",\"itemListElement\":[{\"@type\":\"ListItem\",\"position\":1,\"name\":\"\u9996\u9875\",\"item\":\"https:\/\/www.wsisp.com\/helps\"},{\"@type\":\"ListItem\",\"position\":2,\"name\":\"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b\"}]},{\"@type\":\"WebSite\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#website\",\"url\":\"https:\/\/www.wsisp.com\/helps\/\",\"name\":\"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3\",\"description\":\"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b\",\"potentialAction\":[{\"@type\":\"SearchAction\",\"target\":{\"@type\":\"EntryPoint\",\"urlTemplate\":\"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}\"},\"query-input\":\"required name=search_term_string\"}],\"inLanguage\":\"zh-Hans\"},{\"@type\":\"Person\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41\",\"name\":\"admin\",\"image\":{\"@type\":\"ImageObject\",\"inLanguage\":\"zh-Hans\",\"@id\":\"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/\",\"url\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"contentUrl\":\"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery\",\"caption\":\"admin\"},\"sameAs\":[\"http:\/\/wp.wsisp.com\"],\"url\":\"https:\/\/www.wsisp.com\/helps\/author\/admin\"}]}<\/script>\n<!-- \/ Yoast SEO plugin. -->","yoast_head_json":{"title":"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","robots":{"index":"index","follow":"follow","max-snippet":"max-snippet:-1","max-image-preview":"max-image-preview:large","max-video-preview":"max-video-preview:-1"},"canonical":"https:\/\/www.wsisp.com\/helps\/72684.html","og_locale":"zh_CN","og_type":"article","og_title":"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","og_description":"\u80cc\u666f\u610f\u4e49 \u9f8b\u9f7f\u4f5c\u4e3a\u4e00\u79cd\u5e38\u89c1\u7684\u53e3\u8154\u75be\u75c5，\u5f71\u54cd\u7740\u5168\u7403\u6570\u4ebf\u4eba\u7684\u53e3\u8154\u5065\u5eb7\u3002\u6839\u636e\u4e16\u754c\u536b\u751f\u7ec4\u7ec7\u7684\u7edf\u8ba1，\u9f8b\u9f7f\u7684\u53d1\u75c5\u7387\u5728\u513f\u7ae5\u548c\u9752\u5c11\u5e74\u4e2d\u5c24\u4e3a\u663e\u8457，\u4e14\u968f\u7740\u5e74\u9f84\u7684\u589e\u957f，\u6210\u4eba\u548c\u8001\u5e74\u4eba\u4e5f\u9762\u4e34\u7740\u9f8b\u9f7f\u7684\u5a01\u80c1\u3002\u9f8b\u9f7f\u4e0d\u4ec5\u4f1a\u5bfc\u81f4\u75bc\u75db\u548c\u4e0d\u9002，\u8fd8\u53ef\u80fd\u5f15\u53d1\u66f4\u4e3a\u4e25\u91cd\u7684\u53e3\u8154\u5e76\u53d1\u75c7，\u5982\u7259\u9ad3\u708e\u3001\u6839\u5c16\u5468\u56f4\u708e\u7b49\u3002\u56e0\u6b64，\u65e9\u671f\u7684\u9f8b\u9f7f\u68c0\u6d4b\u4e0e\u8bca\u65ad\u663e\u5f97\u5c24\u4e3a\u91cd\u8981\u3002\u4f20\u7edf\u7684\u9f8b\u9f7f\u8bca\u65ad\u65b9\u6cd5\u4e3b\u8981\u4f9d\u8d56\u4e8e\u533b\u751f\u7684\u89c6\u89c9\u68c0\u67e5\u548cX\u5149\u5f71\u50cf，\u8fd9\u79cd\u65b9\u6cd5\u4e0d\u4ec5\u8017","og_url":"https:\/\/www.wsisp.com\/helps\/72684.html","og_site_name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","article_published_time":"2026-02-06T03:50:01+00:00","og_image":[{"url":"https:\/\/www.wsisp.com\/helps\/wp-content\/uploads\/2026\/02\/20260206034950-698564de8d177.jpg"}],"author":"admin","twitter_card":"summary_large_image","twitter_misc":{"\u4f5c\u8005":"admin","\u9884\u8ba1\u9605\u8bfb\u65f6\u95f4":"13 \u5206"},"schema":{"@context":"https:\/\/schema.org","@graph":[{"@type":"WebPage","@id":"https:\/\/www.wsisp.com\/helps\/72684.html","url":"https:\/\/www.wsisp.com\/helps\/72684.html","name":"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b - \u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","isPartOf":{"@id":"https:\/\/www.wsisp.com\/helps\/#website"},"datePublished":"2026-02-06T03:50:01+00:00","dateModified":"2026-02-06T03:50:01+00:00","author":{"@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41"},"breadcrumb":{"@id":"https:\/\/www.wsisp.com\/helps\/72684.html#breadcrumb"},"inLanguage":"zh-Hans","potentialAction":[{"@type":"ReadAction","target":["https:\/\/www.wsisp.com\/helps\/72684.html"]}]},{"@type":"BreadcrumbList","@id":"https:\/\/www.wsisp.com\/helps\/72684.html#breadcrumb","itemListElement":[{"@type":"ListItem","position":1,"name":"\u9996\u9875","item":"https:\/\/www.wsisp.com\/helps"},{"@type":"ListItem","position":2,"name":"\u3010\u5b8c\u6574\u6e90\u7801+\u6570\u636e\u96c6+\u90e8\u7f72\u6559\u7a0b\u3011\u9f8b\u9f7f\u7259\u9f7f\u75c5\u53d8\u56fe\u50cf\u5206\u5272\u7cfb\u7edf\u6e90\u7801\uff06\u6570\u636e\u96c6\u5206\u4eab [yolov8-seg-C2f-DCNV3\uff06yolov8-seg-RepHGNetV2\u7b4950+\u5168\u5957\u6539\u8fdb\u521b\u65b0\u70b9\u53d1\u520a_\u4e00\u952e\u8bad\u7ec3\u6559\u7a0b"}]},{"@type":"WebSite","@id":"https:\/\/www.wsisp.com\/helps\/#website","url":"https:\/\/www.wsisp.com\/helps\/","name":"\u7f51\u7855\u4e92\u8054\u5e2e\u52a9\u4e2d\u5fc3","description":"\u9999\u6e2f\u670d\u52a1\u5668_\u9999\u6e2f\u4e91\u670d\u52a1\u5668\u8d44\u8baf_\u670d\u52a1\u5668\u5e2e\u52a9\u6587\u6863_\u670d\u52a1\u5668\u6559\u7a0b","potentialAction":[{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https:\/\/www.wsisp.com\/helps\/?s={search_term_string}"},"query-input":"required name=search_term_string"}],"inLanguage":"zh-Hans"},{"@type":"Person","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/358e386c577a3ab51c4493330a20ad41","name":"admin","image":{"@type":"ImageObject","inLanguage":"zh-Hans","@id":"https:\/\/www.wsisp.com\/helps\/#\/schema\/person\/image\/","url":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","contentUrl":"https:\/\/gravatar.wp-china-yes.net\/avatar\/?s=96&d=mystery","caption":"admin"},"sameAs":["http:\/\/wp.wsisp.com"],"url":"https:\/\/www.wsisp.com\/helps\/author\/admin"}]}},"_links":{"self":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/72684","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/comments?post=72684"}],"version-history":[{"count":0,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/posts\/72684\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media\/72675"}],"wp:attachment":[{"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/media?parent=72684"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/categories?post=72684"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/tags?post=72684"},{"taxonomy":"topic","embeddable":true,"href":"https:\/\/www.wsisp.com\/helps\/wp-json\/wp\/v2\/topic?post=72684"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}